bpf: Fix __reg_bound_offset 64->32 var_off subreg propagation

[ Upstream commit 7be14c1c9030f73cc18b4ff23b78a0a081f16188 ]

Xu reports that after commit 3f50f132d840 ("bpf: Verifier, do explicit ALU32
bounds tracking"), the following BPF program is rejected by the verifier:

   0: (61) r2 = *(u32 *)(r1 +0)          ; R2_w=pkt(off=0,r=0,imm=0)
   1: (61) r3 = *(u32 *)(r1 +4)          ; R3_w=pkt_end(off=0,imm=0)
   2: (bf) r1 = r2
   3: (07) r1 += 1
   4: (2d) if r1 > r3 goto pc+8
   5: (71) r1 = *(u8 *)(r2 +0)           ; R1_w=scalar(umax=255,var_off=(0x0; 0xff))
   6: (18) r0 = 0x7fffffffffffff10
   8: (0f) r1 += r0                      ; R1_w=scalar(umin=0x7fffffffffffff10,umax=0x800000000000000f)
   9: (18) r0 = 0x8000000000000000
  11: (07) r0 += 1
  12: (ad) if r0 < r1 goto pc-2
  13: (b7) r0 = 0
  14: (95) exit

And the verifier log says:

  func#0 @0
  0: R1=ctx(off=0,imm=0) R10=fp0
  0: (61) r2 = *(u32 *)(r1 +0)          ; R1=ctx(off=0,imm=0) R2_w=pkt(off=0,r=0,imm=0)
  1: (61) r3 = *(u32 *)(r1 +4)          ; R1=ctx(off=0,imm=0) R3_w=pkt_end(off=0,imm=0)
  2: (bf) r1 = r2                       ; R1_w=pkt(off=0,r=0,imm=0) R2_w=pkt(off=0,r=0,imm=0)
  3: (07) r1 += 1                       ; R1_w=pkt(off=1,r=0,imm=0)
  4: (2d) if r1 > r3 goto pc+8          ; R1_w=pkt(off=1,r=1,imm=0) R3_w=pkt_end(off=0,imm=0)
  5: (71) r1 = *(u8 *)(r2 +0)           ; R1_w=scalar(umax=255,var_off=(0x0; 0xff)) R2_w=pkt(off=0,r=1,imm=0)
  6: (18) r0 = 0x7fffffffffffff10       ; R0_w=9223372036854775568
  8: (0f) r1 += r0                      ; R0_w=9223372036854775568 R1_w=scalar(umin=9223372036854775568,umax=9223372036854775823,s32_min=-240,s32_max=15)
  9: (18) r0 = 0x8000000000000000       ; R0_w=-9223372036854775808
  11: (07) r0 += 1                      ; R0_w=-9223372036854775807
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775807 R1_w=scalar(umin=9223372036854775568,umax=9223372036854775809)
  13: (b7) r0 = 0                       ; R0_w=0
  14: (95) exit

  from 12 to 11: R0_w=-9223372036854775807 R1_w=scalar(umin=9223372036854775810,umax=9223372036854775823,var_off=(0x8000000000000000; 0xffffffff)) R2_w=pkt(off=0,r=1,imm=0) R3_w=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775806
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775806 R1_w=scalar(umin=9223372036854775810,umax=9223372036854775810,var_off=(0x8000000000000000; 0xffffffff))
  13: safe

  [...]

  from 12 to 11: R0_w=-9223372036854775795 R1=scalar(umin=9223372036854775822,umax=9223372036854775823,var_off=(0x8000000000000000; 0xffffffff)) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775794
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775794 R1=scalar(umin=9223372036854775822,umax=9223372036854775822,var_off=(0x8000000000000000; 0xffffffff))
  13: safe

  from 12 to 11: R0_w=-9223372036854775794 R1=scalar(umin=9223372036854775823,umax=9223372036854775823,var_off=(0x8000000000000000; 0xffffffff)) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775793
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775793 R1=scalar(umin=9223372036854775823,umax=9223372036854775823,var_off=(0x8000000000000000; 0xffffffff))
  13: safe

  from 12 to 11: R0_w=-9223372036854775793 R1=scalar(umin=9223372036854775824,umax=9223372036854775823,var_off=(0x8000000000000000; 0xffffffff)) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775792
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775792 R1=scalar(umin=9223372036854775824,umax=9223372036854775823,var_off=(0x8000000000000000; 0xffffffff))
  13: safe

  [...]

The 64bit umin=9223372036854775810 bound continuously bumps by +1 while
umax=9223372036854775823 stays as-is until the verifier complexity limit
is reached and the program gets finally rejected. During this simulation,
the umin also eventually surpasses umax. Looking at the first 'from 12
to 11' output line from the loop, R1 has the following state:

  R1_w=scalar(umin=0x8000000000000002 (9223372036854775810),
              umax=0x800000000000000f (9223372036854775823),
          var_off=(0x8000000000000000;
                           0xffffffff))

The var_off has technically not an inconsistent state but it's very
imprecise and far off surpassing 64bit umax bounds whereas the expected
output with refined known bits in var_off should have been like:

  R1_w=scalar(umin=0x8000000000000002 (9223372036854775810),
              umax=0x800000000000000f (9223372036854775823),
          var_off=(0x8000000000000000;
                                  0xf))

In the above log, var_off stays as var_off=(0x8000000000000000; 0xffffffff)
and does not converge into a narrower mask where more bits become known,
eventually transforming R1 into a constant upon umin=9223372036854775823,
umax=9223372036854775823 case where the verifier would have terminated and
let the program pass.

The __reg_combine_64_into_32() marks the subregister unknown and propagates
64bit {s,u}min/{s,u}max bounds to their 32bit equivalents iff they are within
the 32bit universe. The question came up whether __reg_combine_64_into_32()
should special case the situation that when 64bit {s,u}min bounds have
the same value as 64bit {s,u}max bounds to then assign the latter as
well to the 32bit reg->{s,u}32_{min,max}_value. As can be seen from the
above example however, that is just /one/ special case and not a /generic/
solution given above example would still not be addressed this way and
remain at an imprecise var_off=(0x8000000000000000; 0xffffffff).

The improvement is needed in __reg_bound_offset() to refine var32_off with
the updated var64_off instead of the prior reg->var_off. The reg_bounds_sync()
code first refines information about the register's min/max bounds via
__update_reg_bounds() from the current var_off, then in __reg_deduce_bounds()
from sign bit and with the potentially learned bits from bounds it'll
update the var_off tnum in __reg_bound_offset(). For example, intersecting
with the old var_off might have improved bounds slightly, e.g. if umax
was 0x7f...f and var_off was (0; 0xf...fc), then new var_off will then
result in (0; 0x7f...fc). The intersected var64_off holds then the
universe which is a superset of var32_off. The point for the latter is
not to broaden, but to further refine known bits based on the intersection
of var_off with 32 bit bounds, so that we later construct the final var_off
from upper and lower 32 bits. The final __update_reg_bounds() can then
potentially still slightly refine bounds if more bits became known from the
new var_off.

After the improvement, we can see R1 converging successively:

  func#0 @0
  0: R1=ctx(off=0,imm=0) R10=fp0
  0: (61) r2 = *(u32 *)(r1 +0)          ; R1=ctx(off=0,imm=0) R2_w=pkt(off=0,r=0,imm=0)
  1: (61) r3 = *(u32 *)(r1 +4)          ; R1=ctx(off=0,imm=0) R3_w=pkt_end(off=0,imm=0)
  2: (bf) r1 = r2                       ; R1_w=pkt(off=0,r=0,imm=0) R2_w=pkt(off=0,r=0,imm=0)
  3: (07) r1 += 1                       ; R1_w=pkt(off=1,r=0,imm=0)
  4: (2d) if r1 > r3 goto pc+8          ; R1_w=pkt(off=1,r=1,imm=0) R3_w=pkt_end(off=0,imm=0)
  5: (71) r1 = *(u8 *)(r2 +0)           ; R1_w=scalar(umax=255,var_off=(0x0; 0xff)) R2_w=pkt(off=0,r=1,imm=0)
  6: (18) r0 = 0x7fffffffffffff10       ; R0_w=9223372036854775568
  8: (0f) r1 += r0                      ; R0_w=9223372036854775568 R1_w=scalar(umin=9223372036854775568,umax=9223372036854775823,s32_min=-240,s32_max=15)
  9: (18) r0 = 0x8000000000000000       ; R0_w=-9223372036854775808
  11: (07) r0 += 1                      ; R0_w=-9223372036854775807
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775807 R1_w=scalar(umin=9223372036854775568,umax=9223372036854775809)
  13: (b7) r0 = 0                       ; R0_w=0
  14: (95) exit

  from 12 to 11: R0_w=-9223372036854775807 R1_w=scalar(umin=9223372036854775810,umax=9223372036854775823,var_off=(0x8000000000000000; 0xf),s32_min=0,s32_max=15,u32_max=15) R2_w=pkt(off=0,r=1,imm=0) R3_w=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775806
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775806 R1_w=-9223372036854775806
  13: safe

  from 12 to 11: R0_w=-9223372036854775806 R1_w=scalar(umin=9223372036854775811,umax=9223372036854775823,var_off=(0x8000000000000000; 0xf),s32_min=0,s32_max=15,u32_max=15) R2_w=pkt(off=0,r=1,imm=0) R3_w=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775805
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775805 R1_w=-9223372036854775805
  13: safe

  [...]

  from 12 to 11: R0_w=-9223372036854775798 R1=scalar(umin=9223372036854775819,umax=9223372036854775823,var_off=(0x8000000000000008; 0x7),s32_min=8,s32_max=15,u32_min=8,u32_max=15) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775797
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775797 R1=-9223372036854775797
  13: safe

  from 12 to 11: R0_w=-9223372036854775797 R1=scalar(umin=9223372036854775820,umax=9223372036854775823,var_off=(0x800000000000000c; 0x3),s32_min=12,s32_max=15,u32_min=12,u32_max=15) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775796
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775796 R1=-9223372036854775796
  13: safe

  from 12 to 11: R0_w=-9223372036854775796 R1=scalar(umin=9223372036854775821,umax=9223372036854775823,var_off=(0x800000000000000c; 0x3),s32_min=12,s32_max=15,u32_min=12,u32_max=15) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775795
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775795 R1=-9223372036854775795
  13: safe

  from 12 to 11: R0_w=-9223372036854775795 R1=scalar(umin=9223372036854775822,umax=9223372036854775823,var_off=(0x800000000000000e; 0x1),s32_min=14,s32_max=15,u32_min=14,u32_max=15) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775794
  12: (ad) if r0 < r1 goto pc-2         ; R0_w=-9223372036854775794 R1=-9223372036854775794
  13: safe

  from 12 to 11: R0_w=-9223372036854775794 R1=-9223372036854775793 R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  11: (07) r0 += 1                      ; R0_w=-9223372036854775793
  12: (ad) if r0 < r1 goto pc-2
  last_idx 12 first_idx 12
  parent didn't have regs=1 stack=0 marks: R0_rw=P-9223372036854775801 R1_r=scalar(umin=9223372036854775815,umax=9223372036854775823,var_off=(0x8000000000000000; 0xf),s32_min=0,s32_max=15,u32_max=15) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  last_idx 11 first_idx 11
  regs=1 stack=0 before 11: (07) r0 += 1
  parent didn't have regs=1 stack=0 marks: R0_rw=P-9223372036854775805 R1_rw=scalar(umin=9223372036854775812,umax=9223372036854775823,var_off=(0x8000000000000000; 0xf),s32_min=0,s32_max=15,u32_max=15) R2_w=pkt(off=0,r=1,imm=0) R3_w=pkt_end(off=0,imm=0) R10=fp0
  last_idx 12 first_idx 0
  regs=1 stack=0 before 12: (ad) if r0 < r1 goto pc-2
  regs=1 stack=0 before 11: (07) r0 += 1
  regs=1 stack=0 before 12: (ad) if r0 < r1 goto pc-2
  regs=1 stack=0 before 11: (07) r0 += 1
  regs=1 stack=0 before 12: (ad) if r0 < r1 goto pc-2
  regs=1 stack=0 before 11: (07) r0 += 1
  regs=1 stack=0 before 9: (18) r0 = 0x8000000000000000
  last_idx 12 first_idx 12
  parent didn't have regs=2 stack=0 marks: R0_rw=P-9223372036854775801 R1_r=Pscalar(umin=9223372036854775815,umax=9223372036854775823,var_off=(0x8000000000000000; 0xf),s32_min=0,s32_max=15,u32_max=15) R2=pkt(off=0,r=1,imm=0) R3=pkt_end(off=0,imm=0) R10=fp0
  last_idx 11 first_idx 11
  regs=2 stack=0 before 11: (07) r0 += 1
  parent didn't have regs=2 stack=0 marks: R0_rw=P-9223372036854775805 R1_rw=Pscalar(umin=9223372036854775812,umax=9223372036854775823,var_off=(0x8000000000000000; 0xf),s32_min=0,s32_max=15,u32_max=15) R2_w=pkt(off=0,r=1,imm=0) R3_w=pkt_end(off=0,imm=0) R10=fp0
  last_idx 12 first_idx 0
  regs=2 stack=0 before 12: (ad) if r0 < r1 goto pc-2
  regs=2 stack=0 before 11: (07) r0 += 1
  regs=2 stack=0 before 12: (ad) if r0 < r1 goto pc-2
  regs=2 stack=0 before 11: (07) r0 += 1
  regs=2 stack=0 before 12: (ad) if r0 < r1 goto pc-2
  regs=2 stack=0 before 11: (07) r0 += 1
  regs=2 stack=0 before 9: (18) r0 = 0x8000000000000000
  regs=2 stack=0 before 8: (0f) r1 += r0
  regs=3 stack=0 before 6: (18) r0 = 0x7fffffffffffff10
  regs=2 stack=0 before 5: (71) r1 = *(u8 *)(r2 +0)
  13: safe

  from 4 to 13: safe
  verification time 322 usec
  stack depth 0
  processed 56 insns (limit 1000000) max_states_per_insn 1 total_states 3 peak_states 3 mark_read 1

This also fixes up a test case along with this improvement where we match
on the verifier log. The updated log now has a refined var_off, too.

Fixes: 3f50f132d840 ("bpf: Verifier, do explicit ALU32 bounds tracking")
Reported-by: Xu Kuohai <xukuohai@huaweicloud.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Reviewed-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20230314203424.4015351-2-xukuohai@huaweicloud.com
Link: https://lore.kernel.org/bpf/20230322213056.2470-1-daniel@iogearbox.net
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: keep conntrack reference until IPsecv6 policy checks are done

[ Upstream commit b0e214d212030fe497d4d150bb3474e50ad5d093 ]

Keep the conntrack reference until policy checks have been performed for
IPsec V6 NAT support, just like ipv4.

The reference needs to be dropped before a packet is
queued to avoid having the conntrack module unloadable.

Fixes: 58a317f1061c ("netfilter: ipv6: add IPv6 NAT support")
Signed-off-by: Madhu Koriginja <madhu.koriginja@nxp.com>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: dsa: qca8k: remove assignment of an_enabled in pcs_get_state()

[ Upstream commit 9ef70d0130f282638b28cfce24222f71ada00c9c ]

pcs_get_state() implementations are not supposed to alter an_enabled.
Remove this assignment.

Fixes: b3591c2a3661 ("net: dsa: qca8k: Switch to PHYLINK instead of PHYLIB")
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Link: https://lore.kernel.org/r/E1pdsE5-00Dl2l-8F@rmk-PC.armlinux.org.uk
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

libbpf: Fix ld_imm64 copy logic for ksym in light skeleton.

[ Upstream commit a506d6ce1dd184051037dc9d26c3eb187c9fe625 ]

Unlike normal libbpf the light skeleton 'loader' program is doing
btf_find_by_name_kind() call at run-time to find ksym in the kernel and
populate its {btf_id, btf_obj_fd} pair in ld_imm64 insn. To avoid doing the
search multiple times for the same ksym it remembers the first patched ld_imm64
insn and copies {btf_id, btf_obj_fd} from it into subsequent ld_imm64 insn.
Fix a bug in copying logic, since it may incorrectly clear BPF_PSEUDO_BTF_ID flag.

Also replace always true if (btf_obj_fd >= 0) check with unconditional JMP_JA
to clarify the code.

Fixes: d995816b77eb ("libbpf: Avoid reload of imm for weak, unresolved, repeating ksym")
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20230319203014.55866-1-alexei.starovoitov@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/packet: convert po->auxdata to an atomic flag

[ Upstream commit fd53c297aa7b077ae98a3d3d2d3aa278a1686ba6 ]

po->auxdata can be read while another thread
is changing its value, potentially raising KCSAN splat.

Convert it to PACKET_SOCK_AUXDATA flag.

Fixes: 8dc419447415 ("[PACKET]: Add optional checksum computation for recvmsg")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/packet: convert po->origdev to an atomic flag

[ Upstream commit ee5675ecdf7a4e713ed21d98a70c2871d6ebed01 ]

syzbot/KCAN reported that po->origdev can be read
while another thread is changing its value.

We can avoid this splat by converting this field
to an actual bit.

Following patches will convert remaining 1bit fields.

Fixes: 80feaacb8a64 ("[AF_PACKET]: Add option to return orig_dev to userspace.")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/packet: annotate accesses to po->xmit

[ Upstream commit b9d83ab8a708f23a4001d60e9d8d0b3be3d9f607 ]

po->xmit can be set from setsockopt(PACKET_QDISC_BYPASS),
while read locklessly.

Use READ_ONCE()/WRITE_ONCE() to avoid potential load/store
tearing issues.

Fixes: d346a3fae3ff ("packet: introduce PACKET_QDISC_BYPASS socket option")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

vlan: partially enable SIOCSHWTSTAMP in container

[ Upstream commit 731b73dba359e3ff00517c13aa0daa82b34ff466 ]

Setting timestamp filter was explicitly disabled on vlan devices in
containers because it might affect other processes on the host. But it's
absolutely legit in case when real device is in the same namespace.

Fixes: 873017af7784 ("vlan: disable SIOCSHWTSTAMP in container")
Signed-off-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: pcs: xpcs: remove double-read of link state when using AN

[ Upstream commit ef63461caf427a77a04620d74ba90035a712af9c ]

Phylink does not want the current state of the link when reading the
PCS link state - it wants the latched state. Don't double-read the
MII status register. Phylink will re-read as necessary to capture
transient link-down events as of dbae3388ea9c ("net: phylink: Force
retrigger in case of latched link-fail indicator").

The above referenced commit is a dependency for this change, and thus
this change should not be backported to any kernel that does not
contain the above referenced commit.

Fixes: fcb26bd2b6ca ("net: phy: Add Synopsys DesignWare XPCS MDIO module")
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Remove misleading spec_v1 check on var-offset stack read

[ Upstream commit 082cdc69a4651dd2a77539d69416a359ed1214f5 ]

For every BPF_ADD/SUB involving a pointer, adjust_ptr_min_max_vals()
ensures that the resulting pointer has a constant offset if
bypass_spec_v1 is false. This is ensured by calling sanitize_check_bounds()
which in turn calls check_stack_access_for_ptr_arithmetic(). There,
-EACCESS is returned if the register's offset is not constant, thereby
rejecting the program.

In summary, an unprivileged user must never be able to create stack
pointers with a variable offset. That is also the case, because a
respective check in check_stack_write() is missing. If they were able
to create a variable-offset pointer, users could still use it in a
stack-write operation to trigger unsafe speculative behavior [1].

Because unprivileged users must already be prevented from creating
variable-offset stack pointers, viable options are to either remove
this check (replacing it with a clarifying comment), or to turn it
into a "verifier BUG"-message, also adding a similar check in
check_stack_write() (for consistency, as a second-level defense).
This patch implements the first option to reduce verifier bloat.

This check was introduced by commit 01f810ace9ed ("bpf: Allow
variable-offset stack access") which correctly notes that
"variable-offset reads and writes are disallowed (they were already
disallowed for the indirect access case) because the speculative
execution checking code doesn't support them". However, it does not
further discuss why the check in check_stack_read() is necessary.
The code which made this check obsolete was also introduced in this
commit.

I have compiled ~650 programs from the Linux selftests, Linux samples,
Cilium, and libbpf/examples projects and confirmed that none of these
trigger the check in check_stack_read() [2]. Instead, all of these
programs are, as expected, already rejected when constructing the
variable-offset pointers. Note that the check in
check_stack_access_for_ptr_arithmetic() also prints "off=%d" while the
code removed by this patch does not (the error removed does not appear
in the "verification_error" values). For reproducibility, the
repository linked includes the raw data and scripts used to create
the plot.

  [1] https://arxiv.org/pdf/1807.03757.pdf
  [2] https://gitlab.cs.fau.de/un65esoq/bpf-spectre/-/raw/53dc19fcf459c186613b1156a81504b39c8d49db/data/plots/23-02-26_23-56_bpftool/bpftool/0004-errors.pdf?inline=false

Fixes: 01f810ace9ed ("bpf: Allow variable-offset stack access")
Signed-off-by: Luis Gerhorst <gerhorst@cs.fau.de>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20230315165358.23701-1-gerhorst@cs.fau.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/bpf: Fix a fd leak in an error path in network_helpers.c

[ Upstream commit 226efec2b0efad60d4a6c4b2c3a8710dafc4dc21 ]

In __start_server, it leaks a fd when setsockopt(SO_REUSEPORT) fails.
This patch fixes it.

Fixes: eed92afdd14c ("bpf: selftest: Test batching and bpf_(get|set)sockopt in bpf tcp iter")
Reported-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20230316000726.1016773-2-martin.lau@linux.dev
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath11k: fix deinitialization of firmware resources

[ Upstream commit 5a78ac33e3cb8822da64dd1af196e83664b332b0 ]

Currently, in ath11k_ahb_fw_resources_init(), iommu domain
mapping is done only for the chipsets having fixed firmware
memory. Also, for such chipsets, mapping is done only if it
does not have TrustZone support.

During deinitialization, only if TrustZone support is not there,
iommu is unmapped back. However, for non fixed firmware memory
chipsets, TrustZone support is not there and this makes the
condition check to true and it tries to unmap the memory which
was not mapped during initialization.

This leads to the following trace -

[   83.198790] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000008
[   83.259537] Modules linked in: ath11k_ahb ath11k qmi_helpers
.. snip ..
[   83.280286] pstate: 20000005 (nzCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[   83.287228] pc : __iommu_unmap+0x30/0x140
[   83.293907] lr : iommu_unmap+0x5c/0xa4
[   83.298072] sp : ffff80000b3abad0
.. snip ..
[   83.369175] Call trace:
[   83.376282]  __iommu_unmap+0x30/0x140
[   83.378541]  iommu_unmap+0x5c/0xa4
[   83.382360]  ath11k_ahb_fw_resource_deinit.part.12+0x2c/0xac [ath11k_ahb]
[   83.385666]  ath11k_ahb_free_resources+0x140/0x17c [ath11k_ahb]
[   83.392521]  ath11k_ahb_shutdown+0x34/0x40 [ath11k_ahb]
[   83.398248]  platform_shutdown+0x20/0x2c
[   83.403455]  device_shutdown+0x16c/0x1c4
[   83.407621]  kernel_restart_prepare+0x34/0x3c
[   83.411529]  kernel_restart+0x14/0x74
[   83.415781]  __do_sys_reboot+0x1c4/0x22c
[   83.419427]  __arm64_sys_reboot+0x1c/0x24
[   83.423420]  invoke_syscall+0x44/0xfc
[   83.427326]  el0_svc_common.constprop.3+0xac/0xe8
[   83.430974]  do_el0_svc+0xa0/0xa8
[   83.435659]  el0_svc+0x1c/0x44
[   83.438957]  el0t_64_sync_handler+0x60/0x144
[   83.441910]  el0t_64_sync+0x15c/0x160
[   83.446343] Code: aa0103f4 f9400001 f90027a1 d2800001 (f94006a0)
[   83.449903] ---[ end trace 0000000000000000 ]---

This can be reproduced by probing an AHB chipset which is not
having a fixed memory region. During reboot (or rmmod) trace
can be seen.

Fix this issue by adding a condition check on firmware fixed memory
hw_param as done in the counter initialization function.

Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.7.0.1-01744-QCAHKSWPL_SILICONZ-1

Fixes: f9eec4947add ("ath11k: Add support for targets without trustzone")
Signed-off-by: Aditya Kumar Singh <quic_adisi@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20230309095308.24937-1-quic_adisi@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

scm: fix MSG_CTRUNC setting condition for SO_PASSSEC

[ Upstream commit a02d83f9947d8f71904eda4de046630c3eb6802c ]

Currently, kernel would set MSG_CTRUNC flag if msg_control buffer
wasn't provided and SO_PASSCRED was set or if there was pending SCM_RIGHTS.

For some reason we have no corresponding check for SO_PASSSEC.

In the recvmsg(2) doc we have:
       MSG_CTRUNC
              indicates that some control data was discarded due to lack
              of space in the buffer for ancillary data.

So, we need to set MSG_CTRUNC flag for all types of SCM.

This change can break applications those don't check MSG_CTRUNC flag.

Cc: "David S. Miller" <davem@davemloft.net>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Jakub Kicinski <kuba@kernel.org>
Cc: Paolo Abeni <pabeni@redhat.com>
Cc: Leon Romanovsky <leon@kernel.org>
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Alexander Mikhalitsyn <aleksandr.mikhalitsyn@canonical.com>
v2:
- commit message was rewritten according to Eric's suggestion
Acked-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: qat - fix concurrency issue when device state changes

[ Upstream commit 1bdc85550a2b59bb7f62ead7173134e66dd2d60e ]

The sysfs `state` attribute is not protected against race conditions.
If multiple processes perform a device state transition on the same
device in parallel, unexpected behaviors might occur.

For transitioning the device state, adf_sysfs.c calls the functions
adf_dev_init(), adf_dev_start(), adf_dev_stop() and adf_dev_shutdown()
which are unprotected and interdependent on each other. To perform a
state transition, these functions needs to be called in a specific
order:
  * device up:   adf_dev_init() -> adf_dev_start()
  * device down: adf_dev_stop() -> adf_dev_shutdown()

This change introduces the functions adf_dev_up() and adf_dev_down()
which wrap the state machine functions and protect them with a
per-device lock. These are then used in adf_sysfs.c instead of the
individual state transition functions.

Fixes: 5ee52118ac14 ("crypto: qat - expose device state through sysfs for 4xxx")
Signed-off-by: Shashank Gupta <shashank.gupta@intel.com>
Reviewed-by: Giovanni Cabiddu <giovanni.cabiddu@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: fix precision propagation verbose logging

[ Upstream commit 34f0677e7afd3a292bc1aadda7ce8e35faedb204 ]

Fix wrong order of frame index vs register/slot index in precision
propagation verbose (level 2) output. It's wrong and very confusing as is.

Fixes: 529409ea92d5 ("bpf: propagate precision across all frames, not just the last one")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20230313184017.4083374-1-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: take into account liveness when propagating precision

[ Upstream commit 52c2b005a3c18c565fc70cfd0ca49375f301e952 ]

When doing state comparison, if old state has register that is not
marked as REG_LIVE_READ, then we just skip comparison, regardless what's
the state of corresponing register in current state. This is because not
REG_LIVE_READ register is irrelevant for further program execution and
correctness. All good here.

But when we get to precision propagation, after two states were declared
equivalent, we don't take into account old register's liveness, and thus
attempt to propagate precision for register in current state even if
that register in old state was not REG_LIVE_READ anymore. This is bad,
because register in current state could be anything at all and this
could cause -EFAULT due to internal logic bugs.

Fix by taking into account REG_LIVE_READ liveness mark to keep the logic
in state comparison in sync with precision propagation.

Fixes: a3ce685dd01a ("bpf: fix precision tracking")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20230309224131.57449-1-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtw88: mac: Return the original error from rtw_mac_power_switch()

[ Upstream commit 15c8e267dfa62f207ee1db666c822324e3362b84 ]

rtw_mac_power_switch() calls rtw_pwr_seq_parser() which can return
-EINVAL, -EBUSY or 0. Propagate the original error code instead of
unconditionally returning -EINVAL in case of an error.

Fixes: e3037485c68e ("rtw88: new Realtek 802.11ac driver")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230226221004.138331-3-martin.blumenstingl@googlemail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtw88: mac: Return the original error from rtw_pwr_seq_parser()

[ Upstream commit b7ed9fa2cb76ca7a3c3cd4a6d35748fe1fbda9f6 ]

rtw_pwr_seq_parser() calls rtw_sub_pwr_seq_parser() which can either
return -EBUSY, -EINVAL or 0. Propagate the original error code instead
of unconditionally returning -EBUSY in case of an error.

Fixes: e3037485c68e ("rtw88: new Realtek 802.11ac driver")
Signed-off-by: Martin Blumenstingl <martin.blumenstingl@googlemail.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230226221004.138331-2-martin.blumenstingl@googlemail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

tools: bpftool: Remove invalid \' json escape

[ Upstream commit c679bbd611c08b0559ffae079330bc4e5574696a ]

RFC8259 ("The JavaScript Object Notation (JSON) Data Interchange
Format") only specifies \", \\, \/, \b, \f, \n, \r, and \r as valid
two-character escape sequences. This does not include \', which is not
required in JSON because it exclusively uses double quotes as string
separators.

Solidus (/) may be escaped, but does not have to. Only reverse
solidus (\), double quotes ("), and the control characters have to be
escaped. Therefore, with this fix, bpftool correctly supports all valid
two-character escape sequences (but still does not support characters
that require multi-character escape sequences).

Witout this fix, attempting to load a JSON file generated by bpftool
using Python 3.10.6's default json.load() may fail with the error
"Invalid \escape" if the file contains the invalid escaped single
quote (\').

Fixes: b66e907cfee2 ("tools: bpftool: copy JSON writer from iproute2 repository")
Signed-off-by: Luis Gerhorst <gerhorst@cs.fau.de>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Reviewed-by: Quentin Monnet <quentin@isovalent.com>
Link: https://lore.kernel.org/bpf/20230227150853.16863-1-gerhorst@cs.fau.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath6kl: reduce WARN to dev_dbg() in callback

[ Upstream commit 75c4a8154cb6c7239fb55d5550f481f6765fb83c ]

The warn is triggered on a known race condition, documented in the code above
the test, that is correctly handled.  Using WARN() hinders automated testing.
Reducing severity.

Fixes: de2070fc4aa7 ("ath6kl: Fix kernel panic on continuous driver load/unload")
Reported-and-tested-by: syzbot+555908813b2ea35dae9a@syzkaller.appspotmail.com
Signed-off-by: Oliver Neukum <oneukum@suse.com>
Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru>
Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20230126182431.867984-1-pchelkin@ispras.ru
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: brcmfmac: support CQM RSSI notification with older firmware

[ Upstream commit ec52d77d077529f198fd874c550a26b9cc86a331 ]

Using the BCM4339 firmware from linux-firmware (version "BCM4339/2 wl0:
Sep  5 2019 11:05:52 version 6.37.39.113 (r722271 CY)" from
cypress/cyfmac4339-sdio.bin) the RSSI respose is only 4 bytes, which
results in an error being logged.

It seems that older devices send only the RSSI field and neither SNR nor
noise is included.  Handle this by accepting a 4 byte message and
reading only the RSSI from it.

Fixes: 7dd56ea45a66 ("brcmfmac: add support for CQM RSSI notifications")
Signed-off-by: John Keeping <john@metanate.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20230124104248.2917465-1-john@metanate.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath11k: fix SAC bug on peer addition with sta band migration

[ Upstream commit 60b7d62ba8cdbd073997bff0f1cdae8d844002c0 ]

Fix sleep in atomic context warning detected by Smatch static checker
analyzer.

Following the locking pattern for peer_rhash_add lock tbl_mtx_lock mutex
always even if sta is not transitioning to another band.
This is peer_add function and a more secure locking should not cause
performance regression.

Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.5.0.1-01208-QCAHKSWPL_SILICONZ-1

Fixes: d673cb6fe6c0 ("wifi: ath11k: fix peer addition/deletion error on sta band migration")
Reported-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Christian Marangi <ansuelsmth@gmail.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20230209222622.1751-1-ansuelsmth@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath5k: fix an off by one check in ath5k_eeprom_read_freq_list()

[ Upstream commit 4c856ee12df85aabd437c3836ed9f68d94268358 ]

This loop checks that i < max at the start of loop but then it does
i++ which could put it past the end of the array.  It's harmless to
check again and prevent a potential out of bounds.

Fixes: 1048643ea94d ("ath5k: Clean up eeprom parsing and add missing calibration data")
Signed-off-by: Dan Carpenter <error27@gmail.com>
Reviewed-by: Luis Chamberlain <mcgrof@kernel.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/Y+D9hPQrHfWBJhXz@kili
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath5k: Use platform_get_irq() to get the interrupt

[ Upstream commit 95c95251d0547b46d6571e4fbd51b42865c15a4a ]

As of commit a1a2b7125e10 ("of/platform: Drop static setup of IRQ
resource from DT core"), we need to use platform_get_irq() instead of
platform_get_resource() to get our IRQs because
platform_get_resource() simply won't get them anymore.

This was already fixed in several other Atheros WiFi drivers,
apparently in response to Zeal Robot reports. An example of another
fix is commit 9503a1fc123d ("ath9k: Use platform_get_irq() to get the
interrupt"). ath5k seems to have been missed in this effort, though.

Fixes: a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT core")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20230201084131.v2.2.Ic4f8542b0588d7eb4bc6e322d4af3d2064e84ff0@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath11k: Use platform_get_irq() to get the interrupt

[ Upstream commit f117276638b7600b981b3fe28550823cfbe1ef23 ]

As of commit a1a2b7125e10 ("of/platform: Drop static setup of IRQ
resource from DT core"), we need to use platform_get_irq() instead of
platform_get_resource() to get our IRQs because
platform_get_resource() simply won't get them anymore.

This was already fixed in several other Atheros WiFi drivers,
apparently in response to Zeal Robot reports. An example of another
fix is commit 9503a1fc123d ("ath9k: Use platform_get_irq() to get the
interrupt"). ath11k seems to have been missed in this effort, though.

Without this change, WiFi wasn't coming up on my Qualcomm sc7280-based
hardware. Specifically, "platform_get_resource(pdev, IORESOURCE_IRQ,
i)" was failing even for i=0.

Tested-on: WCN6750 hw1.0 AHB WLAN.MSL.1.0.1-00887-QCAMSLSWPLZ-1

Fixes: a1a2b7125e10 ("of/platform: Drop static setup of IRQ resource from DT core")
Fixes: 00402f49d26f ("ath11k: Add support for WCN6750 device")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Tested-by: Jun Yu <junyuu@chromium.org>
Reviewed-by: Lad Prabhakar <prabhakar.mahadev-lad.rj@bp.renesas.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20230201084131.v2.1.I69cf3d56c97098287fe3a70084ee515098390b70@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath9k: hif_usb: fix memory leak of remain_skbs

[ Upstream commit 7654cc03eb699297130b693ec34e25f77b17c947 ]

hif_dev->remain_skb is allocated and used exclusively in
ath9k_hif_usb_rx_stream(). It is implied that an allocated remain_skb is
processed and subsequently freed (in error paths) only during the next
call of ath9k_hif_usb_rx_stream().

So, if the urbs are deallocated between those two calls due to the device
deinitialization or suspend, it is possible that ath9k_hif_usb_rx_stream()
is not called next time and the allocated remain_skb is leaked. Our local
Syzkaller instance was able to trigger that.

remain_skb makes sense when receiving two consecutive urbs which are
logically linked together, i.e. a specific data field from the first skb
indicates a cached skb to be allocated, memcpy'd with some data and
subsequently processed in the next call to ath9k_hif_usb_rx_stream(). Urbs
deallocation supposedly makes that link irrelevant so we need to free the
cached skb in those cases.

Fix the leak by introducing a function to explicitly free remain_skb (if
it is not NULL) when the rx urbs have been deallocated. remain_skb is NULL
when it has not been allocated at all (hif_dev struct is kzalloced) or
when it has been processed in next call to ath9k_hif_usb_rx_stream().

Found by Linux Verification Center (linuxtesting.org) with Syzkaller.

Fixes: fb9987d0f748 ("ath9k_htc: Support for AR9271 chipset.")
Signed-off-by: Fedor Pchelkin <pchelkin@ispras.ru>
Signed-off-by: Alexey Khoroshilov <khoroshilov@ispras.ru>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20230216192301.171225-1-pchelkin@ispras.ru
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath6kl: minor fix for allocation size

[ Upstream commit 778f83f889e7fca37780d9640fcbd0229ae38eaa ]

Although the "param" pointer occupies more or equal space compared
to "*param", the allocation size should use the size of variable
itself.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: bdcd81707973cf8a ("Add ath6kl cleaned up driver")
Signed-off-by: Alexey V. Vissarionov <gremlin@altlinux.org>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20230117110414.GC12547@altlinux.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/chrome: cros_typec_switch: Add missing fwnode_handle_put()

[ Upstream commit dc70234c408c644505a24362b0f095f713e4697e ]

In cros_typec_register_switches(), we should add fwnode_handle_put()
when break out of the iteration device_for_each_child_node()
as it will automatically increase and decrease the refcounter.

Fixes: affc804c44c8 ("platform/chrome: cros_typec_switch: Add switch driver")
Signed-off-by: Liang He <windhl@126.com>
Link: https://lore.kernel.org/r/20230322041657.1857001-1-windhl@126.com
Signed-off-by: Prashant Malani <pmalani@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

hwmon: (pmbus/fsp-3y) Fix functionality bitmask in FSP-3Y YM-2151E

[ Upstream commit 93822f5161a2dc57a60b95b35b3cb8589f53413e ]

The bit flags in pmbus_driver_info functionality for YM-2151E chip were
joined with a comma operator instead of a bitwise OR. This means that
the last constant PMBUS_HAVE_IIN was not OR-ed with the other
PM_BUS_HAVE_* constants for this page but it initialized the next element
of the func array (which was not accessed from anywhere because of the
number of pages).

However, there is no need for setting PMBUS_HAVE_IIN in the 5Vsb page
because this command does not seem to be paged. Obviously, the device
only has one IIN sensor, so it doesn't make sense to query it again from
the second page.

Fixes: 1734b4135a62 ("hwmon: Add driver for fsp-3y PSUs and PDUs")
Signed-off-by: Jan Kundrát <jan.kundrat@cesnet.cz>
Signed-off-by: Tomáš Pecka <tomas.pecka@cesnet.cz>
Link: https://lore.kernel.org/r/20230420171939.212040-1-tomas.pecka@cesnet.cz
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rpmsg: glink: Propagate TX failures in intentless mode as well

[ Upstream commit 7a68f9fa97357a0f2073c9c31ed4101da4fce93e ]

As support for splitting transmission over several messages using
TX_DATA_CONT was introduced it does not immediately return the return
value of qcom_glink_tx().

The result is that in the intentless case (i.e. intent == NULL), the
code will continue to send all additional chunks. This is wasteful, and
it's possible that the send operation could incorrectly indicate
success, if the last chunk fits in the TX fifo.

Fix the condition.

Fixes: 8956927faed3 ("rpmsg: glink: Add TX_DATA_CONT command while sending")
Reviewed-by: Chris Lew <quic_clew@quicinc.com>
Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230418163018.785524-2-quic_bjorande@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

cpufreq: use correct unit when verify cur freq

[ Upstream commit 44295af5019f1997d038ad2611086a2d1e2af167 ]

cpufreq_verify_current_freq checks() if the frequency returned by
the hardware has a slight delta with the valid frequency value
last set and returns "policy->cur" if the delta is within "1 MHz".
In the comparison, "policy->cur" is in "kHz" but it's compared
against HZ_PER_MHZ. So, the comparison range becomes "1 GHz".

Fix this by comparing against KHZ_PER_MHZ instead of HZ_PER_MHZ.

Fixes: f55ae08c8987 ("cpufreq: Avoid unnecessary frequency updates due to mismatch")
Signed-off-by: Sanjay Chandrashekara <sanjayc@nvidia.com>
[ sumit gupta: Commit message update ]
Signed-off-by: Sumit Gupta <sumitg@nvidia.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: bus: Ensure that notify handlers are not running after removal

[ Upstream commit faae443738c6f0dac9b0d3d11d108f6911a989a9 ]

Currently, acpi_device_remove_notify_handler() may return while the
notify handler being removed is still running which may allow the
module holding that handler to be torn down prematurely.

Address this issue by making acpi_device_remove_notify_handler() wait
for the handling of all the ACPI events in progress to complete before
returning.

Fixes: 5894b0c46e49 ("ACPI / scan: Move bus operations and notification routines to bus.c")
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tick/common: Align tick period with the HZ tick.

[ Upstream commit e9523a0d81899361214d118ad60ef76f0e92f71d ]

With HIGHRES enabled tick_sched_timer() is programmed every jiffy to
expire the timer_list timers. This timer is programmed accurate in
respect to CLOCK_MONOTONIC so that 0 seconds and nanoseconds is the
first tick and the next one is 1000/CONFIG_HZ ms later. For HZ=250 it is
every 4 ms and so based on the current time the next tick can be
computed.

This accuracy broke since the commit mentioned below because the jiffy
based clocksource is initialized with higher accuracy in
read_persistent_wall_and_boot_offset(). This higher accuracy is
inherited during the setup in tick_setup_device(). The timer still fires
every 4ms with HZ=250 but timer is no longer aligned with
CLOCK_MONOTONIC with 0 as it origin but has an offset in the us/ns part
of the timestamp. The offset differs with every boot and makes it
impossible for user land to align with the tick.

Align the tick period with CLOCK_MONOTONIC ensuring that it is always a
multiple of 1000/CONFIG_HZ ms.

Fixes: 857baa87b6422 ("sched/clock: Enable sched clock early")
Reported-by: Gusenleitner Klaus <gus@keba.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/20230406095735.0_14edn3@linutronix.de
Link: https://lore.kernel.org/r/20230418122639.ikgfvu3f@linutronix.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/i915: Make intel_get_crtc_new_encoder() less oopsy

[ Upstream commit 631420b06597a33c72b6dcef78d1c2dea17f452d ]

The point of the WARN was to print something, not oops
straight up. Currently that is precisely what happens
if we can't find the connector for the crtc in the atomic
state. Get the dev pointer from the atomic state instead
of the potentially NULL encoder to avoid that.

Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230413200602.6037-2-ville.syrjala@linux.intel.com
Fixes: 3a47ae201e07 ("drm/i915/display: Make WARN* drm specific where encoder ptr is available")
Reviewed-by: Jani Nikula <jani.nikula@intel.com>
(cherry picked from commit 3b6692357f70498f617ea1b31a0378070a0acf1c)
Signed-off-by: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

debugobject: Prevent init race with static objects

[ Upstream commit 63a759694eed61025713b3e14dd827c8548daadc ]

Statically initialized objects are usually not initialized via the init()
function of the subsystem. They are special cased and the subsystem
provides a function to validate whether an object which is not yet tracked
by debugobjects is statically initialized. This means the object is started
to be tracked on first use, e.g. activation.

This works perfectly fine, unless there are two concurrent operations on
that object. Schspa decoded the problem:

T0                      T1

debug_object_assert_init(addr)
  lock_hash_bucket()
  obj = lookup_object(addr);
  if (!obj) {
   unlock_hash_bucket();
- > preemption
            lock_subsytem_object(addr);
      activate_object(addr)
      lock_hash_bucket();
      obj = lookup_object(addr);
      if (!obj) {
     unlock_hash_bucket();
if (is_static_object(addr))
   init_and_track(addr);
      lock_hash_bucket();
      obj = lookup_object(addr);
      obj->state = ACTIVATED;
      unlock_hash_bucket();

    subsys function modifies content of addr,
    so static object detection does
    not longer work.

    unlock_subsytem_object(addr);

        if (is_static_object(addr)) <- Fails

  debugobject emits a warning and invokes the fixup function which
  reinitializes the already active object in the worst case.

This race exists forever, but was never observed until mod_timer() got a
debug_object_assert_init() added which is outside of the timer base lock
held section right at the beginning of the function to cover the lockless
early exit points too.

Rework the code so that the lookup, the static object check and the
tracking object association happens atomically under the hash bucket
lock. This prevents the issue completely as all callers are serialized on
the hash bucket lock and therefore cannot observe inconsistent state.

Fixes: 3ac7fe5a4aab ("infrastructure to debug (dynamic) objects")
Reported-by: syzbot+5093ba19745994288b53@syzkaller.appspotmail.com
Debugged-by: Schspa Shi <schspa@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Link: https://syzkaller.appspot.com/bug?id=22c8a5938eab640d1c6bcc0e3dc7be519d878462
Link: https://lore.kernel.org/lkml/20230303161906.831686-1-schspa@gmail.com
Link: https://lore.kernel.org/r/87zg7dzgao.ffs@tglx
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: add remove function for decoder platform driver

[ Upstream commit e2a10b3801061d05d3e3415b9b824251451cfd6c ]

Need to disable decoder power when remove decoder hardware driver, adding
remove callback function in the definition of platform driver.

Fixes: c05bada35f01 ("media: mtk-vcodec: Add to support multi hardware decode")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: fix decoder disable pm crash

[ Upstream commit 9d2f13fb47dcab6d094f34ecfd6a879a409722b3 ]

Can't call pm_runtime_disable when the architecture support sub device for
'dev->pm.dev' is NUll, or will get below crash log.

[   10.771551] pc : _raw_spin_lock_irq+0x4c/0xa0
[   10.771556] lr : __pm_runtime_disable+0x30/0x130
[   10.771558] sp : ffffffc01e4cb800
[   10.771559] x29: ffffffc01e4cb800 x28: ffffffdf082108a8
[   10.771563] x27: ffffffc01e4cbd70 x26: ffffff8605df55f0
[   10.771567] x25: 0000000000000002 x24: 0000000000000002
[   10.771570] x23: ffffff85c0dc9c00 x22: 0000000000000001
[   10.771573] x21: 0000000000000001 x20: 0000000000000000
[   10.771577] x19: 00000000000000f4 x18: ffffffdf2e9fbe18
[   10.771580] x17: 0000000000000000 x16: ffffffdf2df13c74
[   10.771583] x15: 00000000000002ea x14: 0000000000000058
[   10.771587] x13: ffffffdf2de1b62c x12: ffffffdf2e9e30e4
[   10.771590] x11: 0000000000000000 x10: 0000000000000001
[   10.771593] x9 : 0000000000000000 x8 : 00000000000000f4
[   10.771596] x7 : 6bff6264632c6264 x6 : 0000000000008000
[   10.771600] x5 : 0080000000000000 x4 : 0000000000000001
[   10.771603] x3 : 0000000000000008 x2 : 0000000000000001
[   10.771608] x1 : 0000000000000000 x0 : 00000000000000f4
[   10.771613] Call trace:
[   10.771617]  _raw_spin_lock_irq+0x4c/0xa0
[   10.771620]  __pm_runtime_disable+0x30/0x130
[   10.771657]  mtk_vcodec_probe+0x69c/0x728 [mtk_vcodec_dec 800cc929d6631f79f9b273254c8db94d0d3500dc]
[   10.771662]  platform_drv_probe+0x9c/0xbc
[   10.771665]  really_probe+0x13c/0x3a0
[   10.771668]  driver_probe_device+0x84/0xc0
[   10.771671]  device_driver_attach+0x54/0x78

Fixes: ba31a5b39400 ("media: mtk-vcodec: Remove mtk_vcodec_release_dec_pm")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf/arm-cmn: Fix port detection for CMN-700

[ Upstream commit 2ad91e44e6b0c7ef1ed151b3bb2242a2144e6085 ]

When the "extra device ports" configuration was first added, the
additional mxp_device_port_connect_info registers were added around the
existing mxp_mesh_port_connect_info registers. What I missed about
CMN-700 is that it shuffled them around to remove this discontinuity.
As such, tweak the definitions and factor out a helper for reading these
registers so we can deal with this discrepancy easily, which does at
least allow nicely tidying up the callsites. With this we can then also
do the nice thing and skip accesses completely rather than relying on
RES0 behaviour where we know the extra registers aren't defined.

Fixes: 23760a014417 ("perf/arm-cmn: Add CMN-700 support")
Reported-by: Jing Zhang <renyu.zj@linux.alibaba.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Link: https://lore.kernel.org/r/71d129241d4d7923cde72a0e5b4c8d2f6084525f.1681295193.git.robin.murphy@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: kgdb: Set PSTATE.SS to 1 to re-enable single-step

[ Upstream commit af6c0bd59f4f3ad5daad2f7b777954b1954551d5 ]

Currently only the first attempt to single-step has any effect. After
that all further stepping remains "stuck" at the same program counter
value.

Refer to the ARM Architecture Reference Manual (ARM DDI 0487E.a) D2.12,
PSTATE.SS=1 should be set at each step before transferring the PE to the
'Active-not-pending' state. The problem here is PSTATE.SS=1 is not set
since the second single-step.

After the first single-step, the PE transferes to the 'Inactive' state,
with PSTATE.SS=0 and MDSCR.SS=1, thus PSTATE.SS won't be set to 1 due to
kernel_active_single_step()=true. Then the PE transferes to the
'Active-pending' state when ERET and returns to the debugger by step
exception.

Before this patch:
==================
Entering kdb (current=0xffff3376039f0000, pid 1) on processor 0 due to Keyboard Entry
[0]kdb>

[0]kdb>
[0]kdb> bp write_sysrq_trigger
Instruction(i) BP #0 at 0xffffa45c13d09290 (write_sysrq_trigger)
    is enabled   addr at ffffa45c13d09290, hardtype=0 installed=0

[0]kdb> go
$ echo h > /proc/sysrq-trigger

Entering kdb (current=0xffff4f7e453f8000, pid 175) on processor 1 due to Breakpoint @ 0xffffad651a309290
[1]kdb> ss

Entering kdb (current=0xffff4f7e453f8000, pid 175) on processor 1 due to SS trap @ 0xffffad651a309294
[1]kdb> ss

Entering kdb (current=0xffff4f7e453f8000, pid 175) on processor 1 due to SS trap @ 0xffffad651a309294
[1]kdb>

After this patch:
=================
Entering kdb (current=0xffff6851c39f0000, pid 1) on processor 0 due to Keyboard Entry
[0]kdb> bp write_sysrq_trigger
Instruction(i) BP #0 at 0xffffc02d2dd09290 (write_sysrq_trigger)
    is enabled   addr at ffffc02d2dd09290, hardtype=0 installed=0

[0]kdb> go
$ echo h > /proc/sysrq-trigger

Entering kdb (current=0xffff6851c53c1840, pid 174) on processor 1 due to Breakpoint @ 0xffffc02d2dd09290
[1]kdb> ss

Entering kdb (current=0xffff6851c53c1840, pid 174) on processor 1 due to SS trap @ 0xffffc02d2dd09294
[1]kdb> ss

Entering kdb (current=0xffff6851c53c1840, pid 174) on processor 1 due to SS trap @ 0xffffc02d2dd09298
[1]kdb> ss

Entering kdb (current=0xffff6851c53c1840, pid 174) on processor 1 due to SS trap @ 0xffffc02d2dd0929c
[1]kdb>

Fixes: 44679a4f142b ("arm64: KGDB: Add step debugging support")
Co-developed-by: Wei Li <liwei391@huawei.com>
Signed-off-by: Wei Li <liwei391@huawei.com>
Signed-off-by: Sumit Garg <sumit.garg@linaro.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Daniel Thompson <daniel.thompson@linaro.org>
Tested-by: Daniel Thompson <daniel.thompson@linaro.org>
Link: https://lore.kernel.org/r/20230202073148.657746-3-sumit.garg@linaro.org
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/ioapic: Don't return 0 from arch_dynirq_lower_bound()

[ Upstream commit 5af507bef93c09a94fb8f058213b489178f4cbe5 ]

arch_dynirq_lower_bound() is invoked by the core interrupt code to
retrieve the lowest possible Linux interrupt number for dynamically
allocated interrupts like MSI.

The x86 implementation uses this to exclude the IO/APIC GSI space.
This works correctly as long as there is an IO/APIC registered, but
returns 0 if not. This has been observed in VMs where the BIOS does
not advertise an IO/APIC.

0 is an invalid interrupt number except for the legacy timer interrupt
on x86. The return value is unchecked in the core code, so it ends up
to allocate interrupt number 0 which is subsequently considered to be
invalid by the caller, e.g. the MSI allocation code.

The function has already a check for 0 in the case that an IO/APIC is
registered, as ioapic_dynirq_base is 0 in case of device tree setups.

Consolidate this and zero check for both ioapic_dynirq_base and gsi_top,
which is used in the case that no IO/APIC is registered.

Fixes: 3e5bedc2c258 ("x86/apic: Fix arch_dynirq_lower_bound() bug for DT enabled machines")
Signed-off-by: Saurabh Sengar <ssengar@linux.microsoft.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/1679988604-20308-1-git-send-email-ssengar@linux.microsoft.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

regulator: stm32-pwr: fix of_iomap leak

[ Upstream commit c4a413e56d16a2ae84e6d8992f215c4dcc7fac20 ]

Smatch reports:
drivers/regulator/stm32-pwr.c:166 stm32_pwr_regulator_probe() warn:
'base' from of_iomap() not released on lines: 151,166.

In stm32_pwr_regulator_probe(), base is not released
when devm_kzalloc() fails to allocate memory or
devm_regulator_register() fails to register a new regulator device,
which may cause a leak.

To fix this issue, replace of_iomap() with
devm_platform_ioremap_resource(). devm_platform_ioremap_resource()
is a specialized function for platform devices.
It allows 'base' to be automatically released whether the probe
function succeeds or fails.

Besides, use IS_ERR(base) instead of !base
as the return value of devm_platform_ioremap_resource()
can either be a pointer to the remapped memory or
an ERR_PTR() encoded error code if the operation fails.

Fixes: dc62f951a6a8 ("regulator: stm32-pwr: Fix return value check in stm32_pwr_regulator_probe()")
Signed-off-by: YAN SHI <m202071378@hust.edu.cn>
Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/oe-kbuild-all/202304111750.o2643eJN-lkp@intel.com/
Reviewed-by: Dongliang Mu <dzm91@hust.edu.cn>
Link: https://lore.kernel.org/r/20230412033529.18890-1-m202071378@hust.edu.cn
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: venus: dec: Fix capture formats enumeration order

[ Upstream commit a9d45ec74c8e68aaafe90191928eddbf79f4644f ]

Commit 9593126dae3e ("media: venus: Add a handling of QC08C compressed
format") and commit cef92b14e653 ("media: venus: Add a handling of QC10C
compressed format") added support for the QC08C and QC10C compressed
formats respectively.

But these also caused a regression, because the new formats where added
at the beginning of the vdec_formats[] array and the vdec_inst_init()
function sets the default format output and capture using fixed indexes
of that array:

static void vdec_inst_init(struct venus_inst *inst)
{
...
inst->fmt_out = &vdec_formats[8];
inst->fmt_cap = &vdec_formats[0];
...
}

Since now V4L2_PIX_FMT_NV12 is not the first entry in the array anymore,
the default capture format is not set to that as it was done before.

Both commits changed the first index to keep inst->fmt_out default format
set to V4L2_PIX_FMT_H264, but did not update the latter to keep .fmt_out
default format set to V4L2_PIX_FMT_NV12.

Rather than updating the index to the current V4L2_PIX_FMT_NV12 position,
let's reorder the entries so that this format is the first entry again.

This would also make VIDIOC_ENUM_FMT report the V4L2_PIX_FMT_NV12 format
with an index 0 as it did before the QC08C and QC10C formats were added.

Fixes: 9593126dae3e ("media: venus: Add a handling of QC08C compressed format")
Fixes: cef92b14e653 ("media: venus: Add a handling of QC10C compressed format")
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Stanimir Varbanov <stanimir.k.varbanov@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: venus: dec: Fix handling of the start cmd

[ Upstream commit 50248ad9f190d527cbd578190ca769729518b703 ]

The decoder driver should clear the last_buffer_dequeued flag of the
capture queue upon receiving V4L2_DEC_CMD_START.

The last_buffer_dequeued flag is set upon receiving EOS (which always
happens upon receiving V4L2_DEC_CMD_STOP).

Without this patch, after issuing the V4L2_DEC_CMD_STOP and
V4L2_DEC_CMD_START, the vb2_dqbuf() function will always fail, even if
the buffers are completed by the hardware.

Fixes: beac82904a87 ("media: venus: make decoder compliant with stateful codec API")

Signed-off-by: Michał Krawczyk <mk@semihalf.com>
Signed-off-by: Stanimir Varbanov <stanimir.k.varbanov@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: rc: gpio-ir-recv: Fix support for wake-up

[ Upstream commit 9c592f8ab114875fdb3b2040f01818e53de44991 ]

The driver was intended from the start to be a wake-up source for the
system, however due to the absence of a suitable call to
device_set_wakeup_capable(), the device_may_wakeup() call used to decide
whether to enable the GPIO interrupt as a wake-up source would never
happen. Lookup the DT standard "wakeup-source" property and call
device_init_wakeup() to ensure the device is flagged as being wakeup
capable.

Reported-by: Matthew Lear <matthew.lear@broadcom.com>
Fixes: fd0f6851eb46 ("[media] rc: Add support for GPIO based IR Receiver driver")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/display: Fix potential null dereference

[ Upstream commit 52f1783ff4146344342422c1cd94fcb4ce39b6fe ]

The adev->dm.dc pointer can be NULL and dereferenced in amdgpu_dm_fini()
without checking.

Add a NULL pointer check before calling dc_dmub_srv_destroy().

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Fixes: 9a71c7d31734 ("drm/amd/display: Register DMUB service with DC")
Signed-off-by: Igor Artemiev <Igor.A.Artemiev@mcst.ru>
Signed-off-by: Hamza Mahfooz <hamza.mahfooz@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: hi846: Fix memleak in hi846_init_controls()

[ Upstream commit 2649c1a20e8e399ee955d0e22192f9992662c3d2 ]

hi846_init_controls doesn't clean the allocated ctrl_hdlr
in case there is a failure, which causes memleak. Add
v4l2_ctrl_handler_free to free the resource properly.

Fixes: e8c0882685f9 ("media: i2c: add driver for the SK Hynix Hi-846 8M pixel camera")
Signed-off-by: Wei Chen <harperchen1110@gmail.com>
Reviewed-by: Martin Kepplinger <martin.kepplinger@puri.sm>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: v4l: async: Return async sub-devices to subnotifier list

[ Upstream commit 5276c9df9c2ab9a43b534bfb56bdb10899cd3a22 ]

When an async notifier is unregistered, the async sub-devices in the
notifier's done list will disappear with the notifier. However this is
currently also done to the sub-notifiers that remain registered. Their
sub-devices only need to be unbound while the async sub-devices themselves
need to be returned to the sub-notifier's waiting list. Do this now.

Fixes: 2cab00bb076b ("media: v4l: async: Allow binding notifiers to sub-devices")
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: rcar_fdp1: Fix refcount leak in probe and remove function

[ Upstream commit c766c90faf93897b77c9c5daa603cffab85ba907 ]

rcar_fcp_get() take reference, which should be balanced with
rcar_fcp_put(). Add missing rcar_fcp_put() in fdp1_remove and
the error paths of fdp1_probe() to fix this.

Fixes: 4710b752e029 ("[media] v4l: Add Renesas R-Car FDP1 Driver")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
Signed-off-by: Laurent Pinchart <laurent.pinchart+renesas@ideasonboard.com>
[hverkuil: resolve merge conflict, remove() is now void]
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: platform: mtk-mdp3: fix potential frame size overflow in mdp_try_fmt_mplane()

[ Upstream commit 4168720753ce6c14c5d3a35302fc2e1841383443 ]

Fix overflow risk when setting certain formats whose frame size exceeds
a RGB24 with 7723x7723 resolution.

For example, a 7723x7724 RGB24 frame:
    1. bpl (byte per line) = 7723 * 3.
    2. Overflow occurs when bpl * 7724 * depth.

Fixes: 61890ccaefaf ("media: platform: mtk-mdp3: add MediaTek MDP3 driver")
Signed-off-by: Moudy Ho <moudy.ho@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: saa7134: fix use after free bug in saa7134_finidev due to race condition

[ Upstream commit 30cf57da176cca80f11df0d9b7f71581fe601389 ]

In saa7134_initdev, it will call saa7134_hwinit1. There are three
function invoking here: saa7134_video_init1, saa7134_ts_init1
and saa7134_vbi_init1.

All of them will init a timer with same function. Take
saa7134_video_init1 as an example. It'll bound &dev->video_q.timeout
with saa7134_buffer_timeout.

In buffer_activate, the timer funtcion is started.

If we remove the module or device which will call saa7134_finidev
to make cleanup, there may be a unfinished work. The
possible sequence is as follows, which will cause a
typical UAF bug.

Fix it by canceling the timer works accordingly before cleanup in
saa7134_finidev.

CPU0                  CPU1

                    |saa7134_buffer_timeout
saa7134_finidev     |
  kfree(dev);       |
                    |
                    | saa7134_buffer_next
                    | //use dev

Fixes: 1e7126b4a86a ("media: saa7134: Convert timers to use timer_setup()")
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: dm1105: Fix use after free bug in dm1105_remove due to race condition

[ Upstream commit 5abda7a16698d4d1f47af1168d8fa2c640116b4a ]

In dm1105_probe, it called dm1105_ir_init and bound
&dm1105->ir.work with dm1105_emit_key.
When it handles IRQ request with dm1105_irq,
it may call schedule_work to start the work.

When we call dm1105_remove to remove the driver, there
may be a sequence as follows:

Fix it by finishing the work before cleanup in dm1105_remove

CPU0                  CPU1

                    |dm1105_emit_key
dm1105_remove      |
  dm1105_ir_exit       |
    rc_unregister_device |
    rc_free_device  |
    rc_dev_release  |
    kfree(dev);     |
                    |
                    | rc_keydown
                    |   //use

Fixes: 34d2f9bf189c ("V4L/DVB: dm1105: use dm1105_dev & dev instead of dm1105dvb")
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/amd: pmc: Move out of BIOS SMN pair for STB init

[ Upstream commit 8d99129eef8f42377b41c1bacee9f8ce806e9f44 ]

The current SMN index used for the driver probe seems to be meant
for the BIOS pair and there are potential concurrency problems that can
occur with an inopportune SMI.

It is been advised to use SMN_INDEX_0 instead of SMN_INDEX_6, which is
what amd_nb.c provides and this function has protections to ensure that
only one caller can use it at a time.

Fixes: 426c0ff27b83 ("platform/x86: amd-pmc: Add support for AMD Smart Trace Buffer")
Co-developed-by: Sanket Goswami <Sanket.Goswami@amd.com>
Signed-off-by: Sanket Goswami <Sanket.Goswami@amd.com>
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
Link: https://lore.kernel.org/r/20230409185348.556161-7-Shyam-sundar.S-k@amd.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/amd: pmc: Utilize SMN index 0 for driver probe

[ Upstream commit 310e782a99c7f16fb533a45d8f9c16defefa5aab ]

The current SMN index used for the driver probe seems to be meant
for the BIOS pair and there are potential concurrency problems that can
occur with an inopportune SMI.

It is been advised to use SMN_INDEX_0 instead of SMN_INDEX_2, which is
what amd_nb.c provides and this function has protections to ensure that
only one caller can use it at a time.

Fixes: 156ec4731cb2 ("platform/x86: amd-pmc: Add AMD platform support for S2Idle")
Co-developed-by: Sanket Goswami <Sanket.Goswami@amd.com>
Signed-off-by: Sanket Goswami <Sanket.Goswami@amd.com>
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
Link: https://lore.kernel.org/r/20230409185348.556161-6-Shyam-sundar.S-k@amd.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/amd: pmc: Move idlemask check into `amd_pmc_idlemask_read`

[ Upstream commit 9217bd1d7699f34a01b26ba14ff38c1714ce1185 ]

The version check requirement for idle mask support actually only
applies to RN/CZN/BRC platforms.

So far no issues have happened because the PMFW version string is
bigger on other supported systems.  This can be reset for any new platform
so move the check to only RN/CZN/BRC case.

Fixes: f6045de1f532 ("platform/x86: amd-pmc: Export Idlemask values based on the APU")
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Link: https://lore.kernel.org/r/20230409185348.556161-5-Shyam-sundar.S-k@amd.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/amd: pmc: Don't dump data after resume from s0i3 on picasso

[ Upstream commit 7abc3618b65304d409d9489d77e4a8f047842fb7 ]

This command isn't supported on Picasso, so guard against running it
to avoid errors like `SMU cmd unknown. err: 0xfe` in the logs.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2449
Fixes: 766205674962 ("platform/x86: amd-pmc: Add support for logging SMU metrics")
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Link: https://lore.kernel.org/r/20230409185348.556161-4-Shyam-sundar.S-k@amd.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/amd: pmc: Hide SMU version and program attributes for Picasso

[ Upstream commit 5ec9ee0d464750d72972d5685edf675824e259a1 ]

As the command to get version isn't supported on Picasso, we shouldn't
be exposing this into sysfs either.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2449
Fixes: 7f1ea75d499a ("platform/x86/amd: pmc: Add sysfs files for SMU")
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Link: https://lore.kernel.org/r/20230409185348.556161-3-Shyam-sundar.S-k@amd.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/amd: pmc: Don't try to read SMU version on Picasso

[ Upstream commit b845772677ea19b8e4c032bc07393ef32de4ee39 ]

Picasso doesn't support the command in the uPEP mailbox to try to
read the SMU version.

Link: https://gitlab.freedesktop.org/drm/amd/-/issues/2449
Fixes: f6045de1f532 ("platform/x86: amd-pmc: Export Idlemask values based on the APU")
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Link: https://lore.kernel.org/r/20230409185348.556161-2-Shyam-sundar.S-k@amd.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86/amd/pmf: Move out of BIOS SMN pair for driver probe

[ Upstream commit aec8298c093f052fc8a86f9411b69b23953b0edb ]

The current SMN index used for the driver probe seems to be meant
for the BIOS pair and there are potential concurrency problems that can
occur with an inopportune SMI.

It is been advised to use SMN_INDEX_0 instead of SMN_INDEX_2, which is
what amd_nb.c provides and this function has protections to ensure that
only one caller can use it at a time.

Fixes: da5ce22df5fe ("platform/x86/amd/pmf: Add support for PMF core layer")
Co-developed-by: Patil Rajesh Reddy <Patil.Reddy@amd.com>
Signed-off-by: Patil Rajesh Reddy <Patil.Reddy@amd.com>
Signed-off-by: Shyam Sundar S K <Shyam-sundar.S-k@amd.com>
Link: https://lore.kernel.org/r/20230406164807.50969-4-Shyam-sundar.S-k@amd.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: rkvdec: fix use after free bug in rkvdec_remove

[ Upstream commit 3228cec23b8b29215e18090c6ba635840190993d ]

In rkvdec_probe, rkvdec->watchdog_work is bound with
rkvdec_watchdog_func. Then rkvdec_vp9_run may
be called to start the work.

If we remove the module which will call rkvdec_remove
 to make cleanup, there may be a unfinished work.
 The possible sequence is as follows, which will
 cause a typical UAF bug.

Fix it by canceling the work before cleanup in rkvdec_remove.

CPU0                  CPU1

                    |rkvdec_watchdog_func
rkvdec_remove       |
 rkvdec_v4l2_cleanup|
  v4l2_m2m_release  |
    kfree(m2m_dev); |
                    |
                    | v4l2_m2m_get_curr_priv
                    |   m2m_dev->curr_ctx //use

Fixes: cd33c830448b ("media: rkvdec: Add the rkvdec driver")
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: cedrus: fix use after free bug in cedrus_remove due to race condition

[ Upstream commit 50d0a7aea4809cef87979d4669911276aa23b71f ]

In cedrus_probe, dev->watchdog_work is bound with cedrus_watchdog function.
In cedrus_device_run, it will started by schedule_delayed_work. If there is
an unfinished work in cedrus_remove, there may be a race condition and
trigger UAF bug.

CPU0                  CPU1

                    |cedrus_watchdog
cedrus_remove       |
  v4l2_m2m_release  |
  kfree(m2m_dev)    |
                    |
                    | v4l2_m2m_get_curr_priv
                    |   m2m_dev //use

Fix it by canceling the worker in cedrus_remove.

Fixes: 7c38a551bda1 ("media: cedrus: Add watchdog for job completion")
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
Acked-by: Jernej Skrabec <jernej.skrabec@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: change lat thread decode error condition

[ Upstream commit 960badda95f10fb0c60f6f64978b19eafa9507a7 ]

If lat thread can't get lat buffer, it should be that current instance
don't be schedulded, the driver can't free the src buffer directly.

Fixes: 7b182b8d9c85 ("media: mediatek: vcodec: Refactor get and put capture buffer flow")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: making sure queue_work successfully

[ Upstream commit 2e0ef56d81cb2569624d288b7e95a8a2734a7c74 ]

Putting core work to work queue using queue_work maybe fail, call
queue_work again when the count of core work in work queue is less
than core_list_cnt, making sure all the buffer in core list can be
scheduled.

Fixes: 365e4ba01df4 ("media: mtk-vcodec: Add work queue for core hardware decode")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: remove unused lat_buf

[ Upstream commit af50b13dd3d7d5dbc1f08add1c462398e926a053 ]

Remove unused lat_buf from core list, or leading to core list access
NULL point.

Fixes: 365e4ba01df4 ("media: mtk-vcodec: Add work queue for core hardware decode")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: add core decode done event

[ Upstream commit d227af847ac2d7d90350124a1b2451e4fc1f050c ]

Need to make sure core decode done before current instance is free.

Fixes: 365e4ba01df4 ("media: mtk-vcodec: Add work queue for core hardware decode")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: move lat_buf to the top of core list

[ Upstream commit 2cfca6c1bf8074175ea7a3b6b47f77ebdef8f701 ]

Current instance will decode done when begin to wait lat buf full,
move the lat_buf of current instance to the top of core list to make
sure current instance's lat_buf will be used firstly.

Fixes: 365e4ba01df4 ("media: mtk-vcodec: Add work queue for core hardware decode")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: using each instance lat_buf count replace core ready list

[ Upstream commit f7a3780cf96925670736582b9a623a2c9ffb4166 ]

Core Hardware decoder depends on each instance lat_buf count,
calling queue_work decode again when the lat_buf count of each instance
isn't zero.

Fixes: 365e4ba01df4 ("media: mtk-vcodec: Add work queue for core hardware decode")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: add params to record lat and core lat_buf count

[ Upstream commit 5bbb6e2ca67477ab41163b32e6b3444faea74a5e ]

Using lat_buf to share decoder information between lat and core work
queue, adding params to record the buf count.

Fixes: 365e4ba01df4 ("media: mtk-vcodec: Add work queue for core hardware decode")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: Force capture queue format to MM21

[ Upstream commit 68c7df527657a9e962da7b5b9c0308557357d8dc ]

While the decoder can produce frames in both MM21 and MT21C formats, only
MM21 is currently supported by userspace tools (like gstreamer and libyuv).
In order to ensure userspace keeps working after the SCP firmware is
updated to support both MM21 and MT21C formats, force the MM21 format for
the capture queue.

This is meant as a stopgap solution while dynamic format switching between
MM21 and MT21C isn't implemented in the driver.

Fixes: 7501edef6b1f ("media: mediatek: vcodec: Different codec using different capture format")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: Nicolas F. R. A. Prado <nfraprado@collabora.com>
Tested-by: Nicolas F. R. A. Prado <nfraprado@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: Make MM21 the default capture format

[ Upstream commit 6d020d81b91af80a977061e82de25cafa4456af5 ]

Given that only the MM21 capture format is supported by userspace tools
(like gstreamer and libyuv), make it the default capture format.

This allows us to force the MM21 format even when a MM21 and MT21C capable
firmware is available (which is needed while dynamic format switching isn't
implemented in the driver), without causing the following regressions on
v4l2-compliance:

        fail: v4l2-test-formats.cpp(478): pixelformat 3132544d (MT21) for buftype 9 not reported by ENUM_FMT
    test VIDIOC_G_FMT: FAIL
        fail: v4l2-test-formats.cpp(478): pixelformat 3132544d (MT21) for buftype 9 not reported by ENUM_FMT
    test VIDIOC_TRY_FMT: FAIL
        fail: v4l2-test-formats.cpp(478): pixelformat 3132544d (MT21) for buftype 9 not reported by ENUM_FMT
    test VIDIOC_S_FMT: FAIL

Fixes: 7501edef6b1f ("media: mediatek: vcodec: Different codec using different capture format")
Signed-off-by: Yunfei Dong <yunfei.dong@mediatek.com>
Reviewed-by: Nicolas F. R. A. Prado <nfraprado@collabora.com>
Tested-by: Nicolas F. R. A. Prado <nfraprado@collabora.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

media: mediatek: vcodec: Use 4K frame size when supported by stateful decoder

[ Upstream commit e25528e1dbe52784ac250071653104a8adc848e2 ]

After commit b018be06f3c7 ("media: mediatek: vcodec: Read max resolution
from dec_capability"), the stateful video decoder driver never really
sets its output frame size to 4K.

Parse the decoder capability reported by the firmware, and update the
output frame size in mtk_init_vdec_params to enable 4K frame size when
available.

Fixes: b018be06f3c7 ("media: mediatek: vcodec: Read max resolution from dec_capability")
Signed-off-by: Pin-yen Lin <treapking@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: sc7280: Rename qspi data12 as data23

[ Upstream commit 14acf21c0d3f7b7298ffcd2e5b5db4a476ec6202 ]

There are 4 qspi data pins: data0, data1, data2, and data3. Currently
we have a shared pin state for data0 and data1 (2 lane config) and a
pin state for data2 and data3 (you'd enable both this and the 2 lane
state for 4 lanes). The second state is obviously misnamed. Fix it.

Fixes: 7720ea001b52 ("arm64: dts: qcom: sc7280: Add QSPI node")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230323102605.2.I4043491bb24b1e92267c5033d76cdb0fe60934da@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: sc7180: Rename qspi data12 as data23

[ Upstream commit d84f8f2687bdc67f20262e822b206419bcfd0038 ]

There are 4 qspi data pins: data0, data1, data2, and data3. Currently
we have a shared pin state for data0 and data1 (2 lane config) and a
pin state for data2 and data3 (you'd enable both this and the 2 lane
state for 4 lanes). The second state is obviously misnamed. Fix it.

Fixes: ba3fc6496366 ("arm64: dts: sc7180: Add qupv3_0 and qupv3_1")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230323102605.1.Ifc1b5be04653f4ab119698a5944bfecded2080d6@changeid
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: msm8994-angler: removed clash with smem_region

[ Upstream commit c85c8a992794dfcd7cea7a41871710c27c5592a6 ]

This fixes memory overlap error:
[    0.000000] reserved@6300000 (0x0000000006300000--0x0000000007000000) overlaps with smem_region@6a00000 (0x0000000006a00000--0x0000000006c00000)

smem_region is the same as in downstream (qcom,smem) [1], therefore
split reserved memory into two sections on either side of smem_region.

Not adding labels as it's not expected to be used.

[1] https://android.googlesource.com/kernel/msm/+/refs/heads/android-msm-angler-3.10-marshmallow-mr1/arch/arm/boot/dts/qcom/msm8994.dtsi#948

Fixes: 380cd3a34b7f ("arm64: dts: msm8994-angler: fix the memory map")

Signed-off-by: Petr Vorel <pvorel@suse.cz>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230131200414.24373-3-pvorel@suse.cz
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: msm8994-angler: Fix cont_splash_mem mapping

[ Upstream commit fe88480a6be92ecbf6f205ff3a7d7e5ded0562dd ]

Angler's cont_splash_mem mapping is shorter in downstream [1],
therefore 380cd3a34b7f was wrong. Obviously also 0e5ded926f2a was wrong
(workaround which fixed booting at the time).

This fixes error:
[    0.000000] memory@3401000 (0x0000000003401000--0x0000000005601000) overlaps with tzapp@4800000 (0x0000000004800000--0x0000000006100000)

[1] https://android.googlesource.com/kernel/msm/+/refs/heads/android-msm-angler-3.10-marshmallow-mr1/arch/arm64/boot/dts/huawei/huawei_msm8994_angler_row_vn1/huawei-fingerprint.dtsi#16

Fixes: 380cd3a34b7f ("arm64: dts: msm8994-angler: fix the memory map")
Fixes: 0e5ded926f2a ("arm64: dts: qcom: msm8994-angler: Disable cont_splash_mem")

Signed-off-by: Petr Vorel <pvorel@suse.cz>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230131200414.24373-2-pvorel@suse.cz
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/apic: Fix atomic update of offset in reserve_eilvt_offset()

[ Upstream commit f96fb2df3eb31ede1b34b0521560967310267750 ]

The detection of atomic update failure in reserve_eilvt_offset() is
not correct. The value returned by atomic_cmpxchg() should be compared
to the old value from the location to be updated.

If these two are the same, then atomic update succeeded and
"eilvt_offsets[offset]" location is updated to "new" in an atomic way.

Otherwise, the atomic update failed and it should be retried with the
value from "eilvt_offsets[offset]" - exactly what atomic_try_cmpxchg()
does in a correct and more optimal way.

Fixes: a68c439b1966c ("apic, x86: Check if EILVT APIC registers are available (AMD only)")
Signed-off-by: Uros Bizjak <ubizjak@gmail.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/20230227160917.107820-1-ubizjak@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

regulator: core: Avoid lockdep reports when resolving supplies

[ Upstream commit cba6cfdc7c3f1516f0d08ddfb24e689af0932573 ]

An automated bot told me that there was a potential lockdep problem
with regulators. This was on the chromeos-5.15 kernel, but I see
nothing that would be different downstream compared to upstream. The
bot said:
  ============================================
  WARNING: possible recursive locking detected
  5.15.104-lockdep-17461-gc1e499ed6604 #1 Not tainted
  --------------------------------------------
  kworker/u16:4/115 is trying to acquire lock:
  ffffff8083110170 (regulator_ww_class_mutex){+.+.}-{3:3}, at: create_regulator+0x398/0x7ec

  but task is already holding lock:
  ffffff808378e170 (regulator_ww_class_mutex){+.+.}-{3:3}, at: ww_mutex_trylock+0x3c/0x7b8

  other info that might help us debug this:
   Possible unsafe locking scenario:

         CPU0
         ----
    lock(regulator_ww_class_mutex);
    lock(regulator_ww_class_mutex);

   *** DEADLOCK ***

   May be due to missing lock nesting notation

  4 locks held by kworker/u16:4/115:
   #0: ffffff808006a948 ((wq_completion)events_unbound){+.+.}-{0:0}, at: process_one_work+0x520/0x1348
   #1: ffffffc00e0a7cc0 ((work_completion)(&entry->work)){+.+.}-{0:0}, at: process_one_work+0x55c/0x1348
   #2: ffffff80828a2260 (&dev->mutex){....}-{3:3}, at: __device_attach_async_helper+0xd0/0x2a4
   #3: ffffff808378e170 (regulator_ww_class_mutex){+.+.}-{3:3}, at: ww_mutex_trylock+0x3c/0x7b8

  stack backtrace:
  CPU: 2 PID: 115 Comm: kworker/u16:4 Not tainted 5.15.104-lockdep-17461-gc1e499ed6604 #1 9292e52fa83c0e23762b2b3aa1bacf5787a4d5da
  Hardware name: Google Quackingstick (rev0+) (DT)
  Workqueue: events_unbound async_run_entry_fn
  Call trace:
   dump_backtrace+0x0/0x4ec
   show_stack+0x34/0x50
   dump_stack_lvl+0xdc/0x11c
   dump_stack+0x1c/0x48
   __lock_acquire+0x16d4/0x6c74
   lock_acquire+0x208/0x750
   __mutex_lock_common+0x11c/0x11f8
   ww_mutex_lock+0xc0/0x440
   create_regulator+0x398/0x7ec
   regulator_resolve_supply+0x654/0x7c4
   regulator_register_resolve_supply+0x30/0x120
   class_for_each_device+0x1b8/0x230
   regulator_register+0x17a4/0x1f40
   devm_regulator_register+0x60/0xd0
   reg_fixed_voltage_probe+0x728/0xaec
   platform_probe+0x150/0x1c8
   really_probe+0x274/0xa20
   __driver_probe_device+0x1dc/0x3f4
   driver_probe_device+0x78/0x1c0
   __device_attach_driver+0x1ac/0x2c8
   bus_for_each_drv+0x11c/0x190
   __device_attach_async_helper+0x1e4/0x2a4
   async_run_entry_fn+0xa0/0x3ac
   process_one_work+0x638/0x1348
   worker_thread+0x4a8/0x9c4
   kthread+0x2e4/0x3a0
   ret_from_fork+0x10/0x20

The problem was first reported soon after we made many of the
regulators probe asynchronously, though nothing I've seen implies that
the problems couldn't have also happened even without that.

I haven't personally been able to reproduce the lockdep issue, but the
issue does look somewhat legitimate. Specifically, it looks like in
regulator_resolve_supply() we are holding a "rdev" lock while calling
set_supply() -> create_regulator() which grabs the lock of a
_different_ "rdev" (the one for our supply). This is not necessarily
safe from a lockdep perspective since there is no documented ordering
between these two locks.

In reality, we should always be locking a regulator before the
supplying regulator, so I don't expect there to be any real deadlocks
in practice. However, the regulator framework in general doesn't
express this to lockdep.

Let's fix the issue by simply grabbing the two locks involved in the
same way we grab multiple locks elsewhere in the regulator framework:
using the "wound/wait" mechanisms.

Fixes: eaa7995c529b ("regulator: core: avoid regulator_resolve_supply() race condition")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20230329143317.RFC.v2.2.I30d8e1ca10cfbe5403884cdd192253a2e063eb9e@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

regulator: core: Consistently set mutex_owner when using ww_mutex_lock_slow()

[ Upstream commit b83a1772be854f87602de14726737d3e5b06e1f4 ]

When a codepath locks a rdev using ww_mutex_lock_slow() directly then
that codepath is responsible for incrementing the "ref_cnt" and also
setting the "mutex_owner" to "current".

The regulator core consistently got that right for "ref_cnt" but
didn't always get it right for "mutex_owner". Let's fix this.

It's unlikely that this truly matters because the "mutex_owner" is
only needed if we're going to do subsequent locking of the same
rdev. However, even though it's not truly needed it seems less
surprising if we consistently set "mutex_owner" properly.

Fixes: f8702f9e4aa7 ("regulator: core: Use ww_mutex for regulators locking")
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20230329143317.RFC.v2.1.I4e9d433ea26360c06dd1381d091c82bb1a4ce843@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/ttm/pool: Fix ttm_pool_alloc error path

[ Upstream commit 379989e7cbdc7aa7496a00ee286ec146c7599cf0 ]

When hitting an error, the error path forgot to unmap dma mappings and
could call set_pages_wb() on already uncached pages.

Fix this by introducing a common ttm_pool_free_range() function that
does the right thing.

v2:
- Simplify that common function (Christian König)
v3:
- Rename that common function to ttm_pool_free_range() (Christian König)

Fixes: d099fc8f540a ("drm/ttm: new TT backend allocation pool v3")
Cc: Christian König <christian.koenig@amd.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Christian Koenig <christian.koenig@amd.com>
Cc: Huang Rui <ray.huang@amd.com>
Cc: dri-devel@lists.freedesktop.org
Signed-off-by: Thomas Hellström <thomas.hellstrom@linux.intel.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230404200650.11043-2-thomas.hellstrom@linux.intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/ttm: optimize pool allocations a bit v2

[ Upstream commit 735c466465eba51deaee3012d8403c10fc7c8c03 ]

If we got a page pool use it as much as possible.

If we can't get more pages from the pool allocate as much as possible.

Only if that still doesn't work reduce the order and try again.

v2: minor cleanups

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221107195808.1873-1-christian.koenig@amd.com
Stable-dep-of: 379989e7cbdc ("drm/ttm/pool: Fix ttm_pool_alloc error path")
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: apq8096-db820c: drop unit address from PMI8994 regulator

[ Upstream commit ec57cbce1a6d9384f8ac1ff966b204dc262f4927 ]

The PMIC regulators are not supposed to have unit addresses.

Fixes: 2317b87a2a6f ("arm64: dts: qcom: db820c: Add vdd_gfx and tie it into mmcc")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230312183622.460488-8-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: msm8994-msft-lumia-octagon: drop unit address from PMI8994 regulator

[ Upstream commit 7a202df0f3eed006e4a9e7c06d62cf67be56c14c ]

The PMIC regulators are not supposed to have unit addresses.

Fixes: 60b214effb80 ("arm64: dts: qcom: msm8994-octagon: Configure regulators")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230312183622.460488-7-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: msm8994-kitakami: drop unit address from PMI8994 regulator

[ Upstream commit 3555dd528ba9c08d6ccd56239c695dbeac3b63e3 ]

The PMIC regulators are not supposed to have unit addresses.

Fixes: e9783584c9b7 ("arm64: dts: qcom: msm8994-kitakami: Add VDD_GFX regulator")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230312183622.460488-6-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: sc7180-trogdor-pazquel: correct trackpad supply

[ Upstream commit 24f39eec6a70768e7c2eb2f3d8158f45050ff75a ]

The hid-over-i2c takes VDD, not VCC supply.  Fix copy-pasta from other
boards which use elan,ekth3000 with valid VCC:

  sc7180-trogdor-pazquel360-lte.dtb: trackpad@15: 'vcc-supply' does not match any of the regexes: 'pinctrl-[0-9]+'

Fixes: fb69f6adaf88 ("arm64: dts: qcom: sc7180: Add pazquel dts files")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230312183622.460488-4-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: sc7180-trogdor-lazor: correct trackpad supply

[ Upstream commit 52e2996f253d82520011340d40dbc1c76ea79208 ]

The hid-over-i2c takes VDD, not VCC supply.  Fix copy-pasta from other
boards which use elan,ekth3000 with valid VCC:

  sc7180-trogdor-lazor-limozeen-nots-r4.dtb: trackpad@2c: 'vcc-supply' does not match any of the regexes: 'pinctrl-[0-9]+'

Fixes: 2c26adb8dbab ("arm64: dts: qcom: Add sc7180-lazor-limozeen skus")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230312183622.460488-3-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: sc7280-herobrine-villager: correct trackpad supply

[ Upstream commit de88b1759b35086d5e63736fb604ea2d06486b1a ]

The hid-over-i2c takes VDD, not VCC supply.  Fix copy-pasta from other
Herobrine boards which use elan,ekth3000 with valid VCC:

  sc7280-herobrine-villager-r1-lte.dtb: trackpad@2c: 'vcc-supply' does not match any of the regexes: 'pinctrl-[0-9]+'

Fixes: ee2a62116015 ("arm64: dts: qcom: sc7280: Add device tree for herobrine villager")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Reviewed-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20230312183622.460488-2-krzysztof.kozlowski@linaro.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

gpu: host1x: Fix memory leak of device names

[ Upstream commit 55879dad0f3ae8468444b42f785ad79eac05fe5b ]

The device names allocated by dev_set_name() need be freed
before module unloading, but they can not be freed because
the kobject's refcount which was set in device_initialize()
has not be decreased to 0.

As comment of device_add() says, if it fails, use only
put_device() drop the refcount, then the name will be
freed in kobejct_cleanup().

device_del() and put_device() can be replaced with
device_unregister(), so call it to unregister the added
successfully devices, and just call put_device() to the
not added device.

Add a release() function to device to avoid null release()
function WARNING in device_release(), it's empty, because
the context devices are freed together in
host1x_memory_context_list_free().

Fixes: 8aa5bcb61612 ("gpu: host1x: Add context device management code")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

gpu: host1x: Fix potential double free if IOMMU is disabled

[ Upstream commit 8466ff24a37a9a18fb935e90dda64f049131ae28 ]

If context device has no IOMMU, the 'cdl->devs' is freed in
error path, but host1x_memory_context_list_init() doesn't
return an error code, so the module can be loaded successfully,
when it's unloading, the host1x_memory_context_list_free() is
called in host1x_remove(), it will cause double free. Set the
'cdl->devs' to NULL after freeing it to avoid double free.

Fixes: 8aa5bcb61612 ("gpu: host1x: Add context device management code")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Reviewed-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

soc: renesas: renesas-soc: Release 'chipid' from ioremap()

[ Upstream commit fc187a46a8e682f0f1167b230792b88de01ceaa0 ]

Smatch reports:

drivers/soc/renesas/renesas-soc.c:536 renesas_soc_init() warn:
'chipid' from ioremap() not released on lines: 475.

If soc_dev_atrr allocation is failed, function renesas_soc_init()
will return without releasing 'chipid' from ioremap().

Fix this by adding function iounmap().

Fixes: cb5508e47e60 ("soc: renesas: Add support for reading product revision for RZ/G2L family")
Signed-off-by: Li Yang <lidaxian@hust.edu.cn>
Reviewed-by: Dan Carpenter <error27@gmail.com>
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Link: https://lore.kernel.org/r/20230331095545.31823-1-lidaxian@hust.edu.cn
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Sasha Levin <sashal@kernel.org>

soc: bcm: brcmstb: biuctrl: fix of_iomap leak

[ Upstream commit c3fbced9af885a6f217fd95509a613d6590916ce ]

Smatch reports:

drivers/soc/bcm/brcmstb/biuctrl.c:291 setup_hifcpubiuctrl_regs() warn:
'cpubiuctrl_base' from of_iomap() not released on lines: 291.

This is because in setup_hifcpubiuctrl_regs(),
cpubiuctrl_base is not released when handle error, which may cause a leak.
To fix this, iounmap is added when handle error.

Fixes: 22f7a9116eba ("soc: brcmstb: Correct CPU_CREDIT_REG offset for Brahma-B53 CPUs")
Signed-off-by: Zhaoyang Li <lizhaoyang04@hust.edu.cn>
Reviewed-by: Dan Carpenter <error27@gmail.com>
Reviewed-by: Dongliang Mu <dzm91@hust.edu.cn>
Link: https://lore.kernel.org/r/20230327115422.1536615-1-lizhaoyang04@hust.edu.cn
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mailbox: mpfs: switch to txdone_poll

[ Upstream commit b5984a9844fc45cd301a28fb56f3de95f7e20f3c ]

The system controller on PolarFire SoC has no interrupt to signify that
the TX has been completed. The interrupt instead signals that a service
requested by the mailbox client has succeeded. If a service fails, there
will be no interrupt delivered.

Switch to polling the busy register to determine whether transmission
has completed.

Fixes: 83d7b1560810 ("mbox: add polarfire soc system controller mailbox")
Acked-by: Jassi Brar <jaswinder.singh@linaro.org>
Tested-by: Valentina Fernandez <valentina.fernandezalanis@microchip.com>
Signed-off-by: Conor Dooley <conor.dooley@microchip.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/mediatek: dp: Change the aux retries times when receiving AUX_DEFER

[ Upstream commit 9243d70e05c5989f84f840612965f96b524da925 ]

DP 1.4a Section 2.8.7.1.5.6.1:
A DP Source device shall retry at least seven times upon receiving
AUX_DEFER before giving up the AUX transaction.

The drm_dp_i2c_do_msg() function in the drm_dp_helper.c file will
judge the status of the msg->reply parameter passed to aux_transfer
for different processing.

Fixes: f70ac097a2cf ("drm/mediatek: Add MT8195 Embedded DisplayPort driver")
Signed-off-by: Xinlei Lee <xinlei.lee@mediatek.com>
Link: https://patchwork.kernel.org/project/linux-mediatek/patch/1680072203-10394-1-git-send-email-xinlei.lee@mediatek.com/
Signed-off-by: Chun-Kuang Hu <chunkuang.hu@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/lima/lima_drv: Add missing unwind goto in lima_pdev_probe()

[ Upstream commit c5647cae2704e58d1c4e5fedbf63f11bca6376c9 ]

Smatch reports:
drivers/gpu/drm/lima/lima_drv.c:396 lima_pdev_probe() warn:
missing unwind goto?

Store return value in err and goto 'err_out0' which has
lima_sched_slab_fini() before returning.

Fixes: a1d2a6339961 ("drm/lima: driver for ARM Mali4xx GPUs")
Signed-off-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Signed-off-by: Qiang Yu <yuq825@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20230314052711.4061652-1-harshit.m.mogalapalli@oracle.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: VIOT: Initialize the correct IOMMU fwspec

[ Upstream commit 47d26684185d09e083669bbbd0c465ab3493a51f ]

When setting up DMA for a PCI device, we need to initialize its
iommu_fwspec with all possible alias RIDs (such as PCI bridges). To do
this we use pci_for_each_dma_alias() which calls
viot_pci_dev_iommu_init(). This function incorrectly initializes the
fwspec of the bridge instead of the device being configured. Fix it by
passing the original device as context to pci_for_each_dma_alias().

Fixes: 3cf485540e7b ("ACPI: Add driver for the VIOT table")
Link: https://lore.kernel.org/all/Y8qzOKm6kvhGWG1T@myrica
Reported-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Tested-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: mediatek: mt8192-asurada: Fix voltage constraint for Vgpu

[ Upstream commit ff4c868ba8df9dcd144ab4943a50adca1cf33ba2 ]

The MT8192 SoC specifies a maximum voltage for the GPU's digital supply
of 0.88V and the GPU OPPs are declaring a maximum voltage of 0.80V.

In order to keep the GPU voltage in the safe range, change the maximum
voltage for mt6315@7's vbuck1 to 0.80V as sending, for any mistake,
1.193V would be catastrophic.

Fixes: 3183cb62b033 ("arm64: dts: mediatek: asurada: Add SPMI regulators")
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Reviewed-by: Chen-yu Tsai <wenst@chromium.org>
Tested-by: Chen-Yu Tsai <wenst@chromium.org>
Link: https://lore.kernel.org/r/20230301095523.428461-12-angelogioacchino.delregno@collabora.com
Signed-off-by: Matthias Brugger <matthias.bgg@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

cpufreq: qcom-cpufreq-hw: Revert adding cpufreq qos

[ Upstream commit e2b47e585931a988c856fd4ba31e1296f749aee3 ]

The OSM/EPSS hardware controls the frequency of each CPU cluster based
on requests from the OS and various throttling events in the system.
While throttling is in effect the related dcvs interrupt will be kept
high. The purpose of the code handling this interrupt is to
continuously report the thermal pressure based on the throttled
frequency.

The reasoning for adding QoS control to this mechanism is not entirely
clear, but the introduction of commit 'c4c0efb06f17 ("cpufreq:
qcom-cpufreq-hw: Add cpufreq qos for LMh")' causes the
scaling_max_frequncy to be set to the throttled frequency. On the next
iteration of polling, the throttled frequency is above or equal to the
newly requested frequency, so the polling is stopped.

With cpufreq limiting the max frequency, the hardware no longer report a
throttling state and no further updates to thermal pressure or qos
state are made.

The result of this is that scaling_max_frequency can only go down, and
the system becomes slower and slower every time a thermal throttling
event is reported by the hardware.

Even if the logic could be improved, there is no reason for software to
limit the max freqency in response to the hardware limiting the max
frequency. At best software will follow the reported hardware state, but
typically it will cause slower backoff of the throttling.

This reverts commit c4c0efb06f17fa4a37ad99e7752b18a5405c76dc.

Fixes: c4c0efb06f17 ("cpufreq: qcom-cpufreq-hw: Add cpufreq qos for LMh")
Reported-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Bjorn Andersson <quic_bjorande@quicinc.com>
Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

cpufreq: mediatek: Raise proc and sram max voltage for MT7622/7623

[ Upstream commit 0883426fd07e39355362e3f2eb9aee1a154dcaf6 ]

During the addition of SRAM voltage tracking for CCI scaling, this
driver got some voltage limits set for the vtrack algorithm: these
were moved to platform data first, then enforced in a later commit
6a17b3876bc8 ("cpufreq: mediatek: Refine mtk_cpufreq_voltage_tracking()")
using these as max values for the regulator_set_voltage() calls.

In this case, the vsram/vproc constraints for MT7622 and MT7623
were supposed to be the same as MT2701 (and a number of other SoCs),
but that turned out to be a mistake because the aforementioned two
SoCs' maximum voltage for both VPROC and VPROC_SRAM is 1.36V.

Fix that by adding new platform data for MT7622/7623 declaring the
right {proc,sram}_max_volt parameter.

Fixes: ead858bd128d ("cpufreq: mediatek: Move voltage limits to platform data")
Fixes: 6a17b3876bc8 ("cpufreq: mediatek: Refine mtk_cpufreq_voltage_tracking()")
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

cpufreq: mediatek: raise proc/sram max voltage for MT8516

[ Upstream commit d3296bb4cafd4bad4a5cf2eeab9d19cc94f9e30e ]

Since the upper boundary of proc/sram voltage of MT8516 is 1300 mV,
which is greater than the value of MT2701 1150 mV, we fix it by adding
the corresponding platform data and specify proc/sram_max_volt to
support MT8516.

Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com>
Fixes: ead858bd128d ("cpufreq: mediatek: Move voltage limits to platform data")
Fixes: 6a17b3876bc8 ("cpufreq: mediatek: Refine mtk_cpufreq_voltage_tracking()")
Reported-by: Nick Hainke <vincent@systemli.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

cpufreq: mediatek: fix KP caused by handler usage after regulator_put/clk_put

[ Upstream commit d51e106240bc755cbe59634b70d567c192b045b2 ]

Any kind of failure in mtk_cpu_dvfs_info_init() will lead to calling
regulator_put() or clk_put() and the KP will occur since the regulator/clk
handlers are used after released in mtk_cpu_dvfs_info_release().

To prevent the usage after regulator_put()/clk_put(), the regulator/clk
handlers are addressed in a way of "Free the Last Thing Style".

Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com>
Fixes: 4b9ceb757bbb ("cpufreq: mediatek: Enable clocks and regulators")
Suggested-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Suggested-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

cpufreq: mediatek: fix passing zero to 'PTR_ERR'

[ Upstream commit d51c63230994f167126d9d8381011b4cb2b0ad22 ]

In order to prevent passing zero to 'PTR_ERR' in
mtk_cpu_dvfs_info_init(), we fix the return value of of_get_cci() using
error pointer by explicitly casting error number.

Signed-off-by: Jia-Wei Chang <jia-wei.chang@mediatek.com>
Fixes: 0daa47325bae ("cpufreq: mediatek: Link CCI device to CPU")
Reported-by: Dan Carpenter <error27@gmail.com>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: apple: t8103: Disable unused PCIe ports

[ Upstream commit a0189fdfb73dac856b8fa9b9f9581e5099c9391f ]

The PCIe ports are unused (without devices) so disable them instead of
removing them.

Fixes: 7c77ab91b33d ("arm64: dts: apple: Add missing M1 (t8103) devices")
Signed-off-by: Janne Grunau <j@jannau.net>
Reviewed-by: Sven Peter <sven@svenpeter.dev>
Signed-off-by: Hector Martin <marcan@marcan.st>
Signed-off-by: Sasha Levin <sashal@kernel.org>