drm/vc4: Rework the encoder retrieval code

Due to a FIFO that cannot be flushed between the pixelvalve and the HDMI
controllers on BCM2711, we need to carefully disable both at boot time
if they were left enabled by the firmware.

However, at the time we're running that code, the struct drm_connector
encoder pointer isn't set yet, and thus we cannot retrieve the encoder
associated to our CRTC.

We can however make use of the fact that we have a less flexible setup
than what DRM allows where we have a 1:1 relationship between our CRTCs
and encoders (and connectors), and thus store the crtc associated to our
encoder at boot time.

We cannot do that at the time the encoders are probed though, since the
CRTCs won't be probed yet and thus we don't know at that time which CRTC
index we're going to get, so let's do this in two passes: we can first
bind all the components and then once they all are bound, we can iterate
over all the encoders to find their associated CRTC and set the pointer.

This is similar to what we're doing to set the possible_crtcs field.

Fixes: 875a4d536842 ("drm/vc4: drv: Disable the CRTC at boot time")
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

drm/vc4: crtc: Skip the TXP

The vc4_set_crtc_possible_masks is meant to run over all the encoders
and then set their possible_crtcs mask to their associated pixelvalve.

However, since the commit 39fcb2808376 ("drm/vc4: txp: Turn the TXP into
a CRTC of its own"), the TXP has been turned to a CRTC and encoder of
its own, and while it does indeed register an encoder, it no longer has
an associated pixelvalve. The code will thus run over the TXP encoder
and set a bogus possible_crtcs mask, overriding the one set in the TXP
bind function.

In order to fix this, let's skip any virtual encoder.

Cc: <stable@vger.kernel.org> # v5.9+
Fixes: 39fcb2808376 ("drm/vc4: txp: Turn the TXP into a CRTC of its own")
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

drm/vc4: txp: Properly set the possible_crtcs mask

The current code does a binary OR on the possible_crtcs variable of the
TXP encoder, while we want to set it to that value instead.

Cc: <stable@vger.kernel.org> # v5.9+
Fixes: 39fcb2808376 ("drm/vc4: txp: Turn the TXP into a CRTC of its own")
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: requests: Ignore if the pointer is null

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

dtoverlays: Add overlays for JDI LT070ME05000 1200x1920 DSI panel

Credit to forum member gizmomouse on
https://www.raspberrypi.org/forums/viewtopic.php?f=98&t=253912 and
Andrey Vostrukhin of Harlab for these overlays.

See https://github.com/harlab/CM4_LCD_LT070ME05000 for
schematics and docs for the adapter board to connect this panel which
is found in the Asus/Google 2013 Nexus 7" tablet and therefore
relatively easily available.

Note that this uses 4 DSI data lanes, and therefore MUST be used
with DISP1 on a Compute Module. It can not be used on a standard
Pi.

There are two versions of the adapter board. V1 connects the
display controls to Pi GPIOs, whilst v2 uses an I2C GPIO expander
so needs no additional connections beyond the FFC and power.

The touchscreen overlay for these panels varies, so that part
is not configured.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

defconfigs: Add PANEL_JDI_LT070ME05000 DSI panel

Used by the 2013 Nexus 7" tablet, and working via an adapter
board on Compute Modules.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

drm/panel: jdi-lt070me05000: Use gpiod_set_value_cansleep

There is no reason why the control GPIOs for the panel can not
be connected to I2C or similar GPIO interfaces that may need to
sleep, therefore switch from gpiod_set_value to
gpiod_set_value_cansleep calls to configure them.
Without that you get complaints from gpiolib every time the state
is changed.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

drm/atomic: Don't fixup modes that haven't been reset

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

drm/vc4: Add correct stop condition to vc4_dsi_encoder_disable iteration

vc4_dsi_encoder_disable is partially an open coded version of
drm_bridge_chain_disable, but it missed a termination condition
in the loop for ->disable which meant that no post_disable
calls were made.

Add in the termination clause.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

drm/vc4: Fix dsi0 interrupt support.

DSI0 seemingly had very little or no testing as a load of
the register mappings were incorrect/missing, so host
transfers always timed out due to enabling/checking incorrect
bits in the interrupt enable and status registers.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

drm/vc4: Register dsi0 as the correct vc4 encoder type

vc4_dsi was registering both dsi0 and dsi1 as VC4_ENCODER_TYPE_DSI1
which seemed to work OK for a single DSI display, but fails
if there are two DSI displays connected.

Update to register the correct type.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

drm/vc4: Correct pixel order for DSI0

For slightly unknown reasons, dsi0 takes a different pixel format
to dsi1, and that has to be set in the pixel valve.

Amend the setup accordingly.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

drm/vc4: dsi: Add configuration for BCM2711 DSI1

Commit d0666be8ef9e8e65d4b7fabc1606ec51f61384c0 upstream.

BCM2711 DSI1 doesn't have the issue with the ARM not being
able to write to the registers, therefore remove the DMA
workaround for that compatible string.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20201203132543.861591-8-maxime@cerno.tech

drm/vc4: dsi: Add support for DSI0

Commit 4b265fe11fad4234b12d92dd8091f9aa0c878eea upstream.

DSI0 was partially supported, but didn't register with the main
driver, and the code was inconsistent as to whether it checked
port == 0 or port == 1.

Add compatible string and other support to make it consistent.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20201203132543.861591-6-maxime@cerno.tech

drm/vc4: dsi: Introduce a variant structure

Commit d1d195ce26a14ec0a87816c09ae514e1c40e97f7 upstream.

Most of the differences between DSI0 and DSI1 are handled through the
ID. However, the BCM2711 DSI is going to introduce one more variable to
the mix and will break some expectations of the earlier, simpler, test.

Let's add a variant structure that will address most of the differences
between those three controllers.

Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20201203132543.861591-5-maxime@cerno.tech

drm/vc4: dsi: Use snprintf for the PHY clocks instead of an array

Commit dc0bf36401e891c853e0a25baeb4e0b4e6f3626d upstream.

The DSI clocks setup function has been using an array to store the clock
name of either the DSI0 or DSI1 blocks, using the port ID to choose the
proper one.

Let's switch to an snprintf call to do the same thing and simplify the
array a bit.

Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20201203132543.861591-4-maxime@cerno.tech

drm/vc4: drv: Remove the DSI pointer in vc4_drv

Commit 51f4fcd9c4ea867c3b4fe58111f342ad0e80642a upstream.

That pointer isn't used anywhere, so there's no point in keeping it.

Reviewed-by: Dave Stevenson <dave.stevenson@raspberrypi.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20201203132543.861591-2-maxime@cerno.tech

Revert "drm/vc4: Add support for DSI0"

This reverts commit d704a3c965783ed51476031206b2120676418672.

Revert in order to apply upstream version

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

Revert "drm/vc4: Add configuration for BCM2711 DSI1."

This reverts commit 10faa28e1474e6cf33cb0809d2a6d5b7a8351a28.

Revert in order to take the cleaner upstream version.

Signed-off-by: Dave Stevenson <dave.stevenson@raspberrypi.com>

dwc_otg: fix an undeclared variable
Replace an undeclared variable used by DWC_DEBUGPL with the real endpoint address. DWC_DEBUGPL does nothing with DEBUG undefined so it did not go wrong before.
Signed-off-by: Zixuan Wang <wangzixuan@sjtu.edu.cn>

rpivid: Only clk_request_done once

Fixes: 25486f49bfe2e3ae13b90478d1eebd91413136ad
Signed-off-by: Dom Cobley <popcornmix@gmail.com>

r8152: search the configuration of vendor mode

commit c2198943e33b100ed21dfb636c8fa6baef841e9d upstream.

The vendor mode is not always at config #1, so it is necessary to
set the correct configuration number.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: support PHY firmware for RTL8156 series

commit 4a51b0e8a0143b0e83d51d9c58c6416c3818a9f2 upstream.

Support new firmware type and method for RTL8156 series.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: support new chips

commit 195aae321c829dd1945900d75561e6aa79cce208 upstream.

Support RTL8153C, RTL8153D, RTL8156A, and RTL8156B. The RTL8156A
and RTL8156B are the 2.5G ethernet.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: add help function to change mtu

commit 67ce1a806f164e59a074fea8809725d3411eaa20 upstream.

The different chips may have different requests when changing mtu.
Therefore, add a new help function of rtl_ops to change mtu. Besides,
reset the tx/rx after changing mtu.

Additionally, add mtu_to_size() and size_to_mtu() macros to simplify
the code.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: adjust rtl8152_check_firmware function

commit a8a7be178e81a3d4b6972cbeb0ccd091ca2f9f89 upstream.

Use bits operations to record and check the firmware.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: set inter fram gap time depending on speed

commit 5133bcc7481528e36fff0a3b056601efb704fb32 upstream.

Set the maximum inter frame gap time (144ns) for speed 10M/half and
100M/half. It improves the performance for those speeds. And, there
is no effect for the other speeds.

For 10M/half and 100M/half, the fast inter frame gap time let the
device couldn't use the feature of the aggregation effectively,
because the transfer would be completed fastly. Therefore, use the
maximum value to improve the effect of the aggregation. However, you
may not feel the improvement for fast CPUs, because they compensate
for the effect of the aggregation.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

r8152: spilt rtl_set_eee_plus and r8153b_green_en

commit 40fa7568ac230446d888b7ad402cff9e20fe3ad5 upstream.

Add rtl_eee_plus_en() and rtl_green_en().

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

r8152: replace netif_err with dev_err

commit 156c3207611262266f0eea589ac3f00c5657320e upstream.

Some messages are before calling register_netdev(), so replace
netif_err() with dev_err().

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

r8152: check if the pointer of the function exists

commit c79515e47935c747282c6ed2ee5b2ef039756eeb upstream.

Return error code if autosuspend_en, eee_get, or eee_set don't exist.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

r8152: enable U1/U2 for USB_SPEED_SUPER

commit 7a0ae61acde2cebd69665837170405eced86a6c7 upstream.

U1/U2 shoued be enabled for USB 3.0 or later. The USB 2.0 doesn't
support it.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

r8152: adjust the flow of power cut for RTL8153B

commit 80fd850b31f09263ad175b2f640d5c5c6f76ed41 upstream.

For runtime resuming, the RTL8153B may be resumed from the state
of power cut, when enabling the feature of UPS. Then, the PHY
would be reset, so it is necessary to be initailized again.

Besides, the USB_U1U2_TIMER also has to be set again, so I move
it from r8153b_init() to r8153b_hw_phy_cfg().

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

r8152: replace several functions about phy patch request

commit a08c0d309d8c078d22717d815cf9853f6f2c07bd upstream.

Replace r8153_patch_request() with rtl_phy_patch_request().
Replace r8153_pre_ram_code() with rtl_pre_ram_code().
Replace r8153_post_ram_code() with rtl_post_ram_code().
Add rtl_patch_key_set().

The new functions have an additional parameter. It is used to wait
the patch request command finished. When the PHY is resumed from
the state of power cut, the PHY is at a safe mode and the
OCP_PHY_PATCH_STAT wouldn't be updated. For this situation, it is
safe to set patch request command without waiting OCP_PHY_PATCH_STAT.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: r8152: use new tasklet API

commit f3163f1cb87141c7a41a15a5d4c98b353f807b04 upstream.

This converts the driver to use the new tasklet API introduced in
commit 12cc923f1ccc ("tasklet: Introduce new initialization API")

Signed-off-by: Emil Renner Berthing <kernel@esmil.dk>
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net/usb/r8153_ecm: support ECM mode for RTL8153

commit c1aedf015ebdd0232757a66e2daccf1246bd609c upstream.

Support ECM mode based on cdc_ether with relative mii functions,
when CONFIG_USB_RTL8152 is not set, or the device is not supported
by r8152 driver.

Both r8152 and r8153_ecm would check the return value of
rtl8152_get_version() in porbe(). If rtl8152_get_version()
return none zero value, the r8152 is used for the device
with vendor mode. Otherwise, the r8153_ecm is used for the
device with ECM mode.

Signed-off-by: Hayes Wang <hayeswang@realtek.com>
Link: https://lore.kernel.org/r/1394712342-15778-392-Taiwan-albertk@realtek.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: r8152: Fix a couple of spelling errors in fw_phy_nc's docs

commit 9f07814d01ad085b2d9f1d55b4ce532fb2c27110 upstream.

Fixes the following W=1 kernel build warning(s):

 drivers/net/usb/r8152.c:992: warning: Function parameter or member 'mode_pre' not described in 'fw_phy_nc'
 drivers/net/usb/r8152.c:992: warning: Function parameter or member 'mode_post' not described in 'fw_phy_nc'

Signed-off-by: Lee Jones <lee.jones@linaro.org>
Acked-by: Hayes Wang <hayeswang@realtek.com>
Link: https://lore.kernel.org/r/20201102114512.1062724-28-lee.jones@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

net: usb: r8152: Provide missing documentation for some struct members

commit 34e653efb602e0651867fb5ab14369b555a61dcd upstream.

Fixes the following W=1 kernel build warning(s):

 drivers/net/usb/r8152.c:934: warning: Function parameter or member 'blk_hdr' not described in 'fw_mac'
 drivers/net/usb/r8152.c:934: warning: Function parameter or member 'reserved' not described in 'fw_mac'
 drivers/net/usb/r8152.c:947: warning: Function parameter or member 'blk_hdr' not described in 'fw_phy_patch_key'
 drivers/net/usb/r8152.c:947: warning: Function parameter or member 'reserved' not described in 'fw_phy_patch_key'
 drivers/net/usb/r8152.c:986: warning: Function parameter or member 'blk_hdr' not described in 'fw_phy_nc'
 drivers/net/usb/r8152.c:986: warning: Function parameter or member 'mode_pre' not described in 'fw_phy_nc'
 drivers/net/usb/r8152.c:986: warning: Function parameter or member 'mode_post' not described in 'fw_phy_nc'
 drivers/net/usb/r8152.c:986: warning: Function parameter or member 'reserved' not described in 'fw_phy_nc'

Signed-off-by: Lee Jones <lee.jones@linaro.org>
Acked-by: Hayes Wang <hayeswang@realtek.com>
Link: https://lore.kernel.org/r/20201102114512.1062724-23-lee.jones@linaro.org
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

rpivid: Switch to new clock api

Signed-off-by: Dom Cobley <popcornmix@gmail.com>

bcm2835-unicam: Switch to new clock api

Signed-off-by: Dom Cobley <popcornmix@gmail.com>

drm/vc4: hdmi: Convert to the new clock request API

The new clock request API allows us to increase the rate of the
core clock as required during mode set while decreasing it when
we're done, resulting in a better power-efficiency.

Signed-off-by: Dom Cobley <popcornmix@gmail.com>

drm/vc4: hdmi: Convert to the new clock request API

The new clock request API allows us to increase the rate of the HSM
clock to match our pixel rate requirements while decreasing it when
we're done, resulting in a better power-efficiency.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

clk: Introduce a clock request API

It's not unusual to find clocks being shared across multiple devices
that need to change the rate depending on what the device is doing at a
given time.

The SoC found on the RaspberryPi4 (BCM2711) is in such a situation
between its two HDMI controllers that share a clock that needs to be
raised depending on the output resolution of each controller.

The current clk_set_rate API doesn't really allow to support that case
since there's really no synchronisation between multiple users, it's
essentially a fire-and-forget solution.

clk_set_min_rate does allow for such a synchronisation, but has another
drawback: it doesn't allow to reduce the clock rate once the work is
over.

In our previous example, this means that if we were to raise the
resolution of one HDMI controller to the largest resolution and then
changing for a smaller one, we would still have the clock running at the
largest resolution rate resulting in a poor power-efficiency.

In order to address both issues, let's create an API that allows user to
create temporary requests to increase the rate to a minimum, before
going back to the initial rate once the request is done.

This introduces mainly two side-effects:

  * There's an interaction between clk_set_rate and requests. This has
    been addressed by having clk_set_rate increasing the rate if it's
    greater than what the requests asked for, and in any case changing
    the rate the clock will return to once all the requests are done.

  * Similarly, clk_round_rate has been adjusted to take the requests
    into account and return a rate that will be greater or equal to the
    requested rates.

Signed-off-by: Maxime Ripard <maxime@cerno.tech>

Merge remote-tracking branch 'stable/linux-5.10.y' into rpi-5.10.y

overlays: ghost-amp: Add DAC mute control

Signed-off-by: Phil Elwell <phil@raspberrypi.com>

sound/usb: call usb_autopm_get_interface() for devices that should not
be suspended

Webcams with microphones are composite devices, and autosuspend is set
at the device level. If uvcvideo is probed after snd-usb-audio, the effect
of the quirk applied by snd-usb-audio is undone by uvcvideo's global
application of autosuspend.

Incrementing the interface's PM refcount in such cases prevents runtime PM
from happening, thus the device is left active.

Signed-off-by: Jonathan Bell <jonathan@raspberrypi.com>

sound/usb: add device quirks for A4Tech FHD 1080p webcams

These devices use a type of Sonix chipset that produces broken microphone
data if suspended/resumed.

They also don't support readback of the sample rate.

Signed-off-by: Jonathan Bell <jonathan@raspberrypi.com>

Linux 5.10.31

Tested-by: Salvatore Bonaccorso <carnil@debian.org>
Tested-by: Pavel Machek (CIP) <pavel@denx.de>
Tested-by: Andrei Rabusov <a.rabusov@tum.de>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Jason Self <jason@bluehome.net>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Link: https://lore.kernel.org/r/20210415144413.165663182@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

xen/events: fix setting irq affinity

The backport of upstream patch 25da4618af240fbec61 ("xen/events: don't
unmask an event channel when an eoi is pending") introduced a
regression for stable kernels 5.10 and older: setting IRQ affinity for
IRQs related to interdomain events would no longer work, as moving the
IRQ to its new cpu was not included in the irq_ack callback for those
events.

Fix that by adding the needed call.

Note that kernels 5.11 and later don't need the explicit moving of the
IRQ to the target cpu in the irq_ack callback, due to a rework of the
affinity setting in kernel 5.11.

Signed-off-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: sfp: cope with SFPs that set both LOS normal and LOS inverted

commit 624407d2cf14ff58e53bf4b2af9595c4f21d606e upstream.

The SFP MSA defines two option bits in byte 65 to indicate how the
Rx_LOS signal on SFP pin 8 behaves:

bit 2 - Loss of Signal implemented, signal inverted from standard
        definition in SFP MSA (often called "Signal Detect").
bit 1 - Loss of Signal implemented, signal as defined in SFP MSA
        (often called "Rx_LOS").

Clearly, setting both bits results in a meaningless situation: it would
mean that LOS is implemented in both the normal sense (1 = signal loss)
and inverted sense (0 = signal loss).

Unfortunately, there are modules out there which set both bits, which
will be initially interpret as "inverted" sense, and then, if the LOS
signal changes state, we will toggle between LINK_UP and WAIT_LOS
states.

Change our LOS handling to give well defined behaviour: only interpret
these bits as meaningful if exactly one is set, otherwise treat it as
if LOS is not implemented.

Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/E1kyYQa-0004iR-CU@rmk-PC.armlinux.org.uk
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Cc: Pali Rohár <pali@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: sfp: relax bitrate-derived mode check

commit 7a77233ec6d114322e2c4f71b4e26dbecd9ea8a7 upstream.

Do not check the encoding when deriving 1000BASE-X from the bitrate
when no other modes are discovered. Some GPON modules (VSOL V2801F
and CarlitoxxPro CPGOS03-0490 v2.0) indicate NRZ encoding with a
1200Mbaud bitrate, but should be driven with 1000BASE-X on the host
side.

Tested-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Russell King <rmk+kernel@armlinux.org.uk>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf map: Tighten snprintf() string precision to pass gcc check on some 32-bit arches

commit 77d02bd00cea9f1a87afe58113fa75b983d6c23a upstream.

Noticed on a debian:experimental mips and mipsel cross build build
environment:

  perfbuilder@ec265a086e9b:~$ mips-linux-gnu-gcc --version | head -1
  mips-linux-gnu-gcc (Debian 10.2.1-3) 10.2.1 20201224
  perfbuilder@ec265a086e9b:~$

    CC       /tmp/build/perf/util/map.o
  util/map.c: In function 'map__new':
  util/map.c:109:5: error: '%s' directive output may be truncated writing between 1 and 2147483645 bytes into a region of size 4096 [-Werror=format-truncation=]
    109 |    "%s/platforms/%s/arch-%s/usr/lib/%s",
        |     ^~
  In file included from /usr/mips-linux-gnu/include/stdio.h:867,
                   from util/symbol.h:11,
                   from util/map.c:2:
  /usr/mips-linux-gnu/include/bits/stdio2.h:67:10: note: '__builtin___snprintf_chk' output 32 or more bytes (assuming 4294967321) into a destination of size 4096
     67 |   return __builtin___snprintf_chk (__s, __n, __USE_FORTIFY_LEVEL - 1,
        |          ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
     68 |        __bos (__s), __fmt, __va_arg_pack ());
        |        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
  cc1: all warnings being treated as errors

Since we have the lenghts for what lands in that place, use it to give
the compiler more info and make it happy.

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Anders Roxell <anders.roxell@linaro.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

netfilter: x_tables: fix compat match/target pad out-of-bound write

commit b29c457a6511435960115c0f548c4360d5f4801d upstream.

xt_compat_match/target_from_user doesn't check that zeroing the area
to start of next rule won't write past end of allocated ruleset blob.

Remove this code and zero the entire blob beforehand.

Reported-by: syzbot+cfc0247ac173f597aaaa@syzkaller.appspotmail.com
Reported-by: Andy Nguyen <theflow@google.com>
Fixes: 9fa492cdc160c ("[NETFILTER]: x_tables: simplify compat API")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

block: don't ignore REQ_NOWAIT for direct IO

[ Upstream commit f8b78caf21d5bc3fcfc40c18898f9d52ed1451a5 ]

If IOCB_NOWAIT is set on submission, then that needs to get propagated to
REQ_NOWAIT on the block side. Otherwise we completely lose this
information, and any issuer of IOCB_NOWAIT IO will potentially end up
blocking on eg request allocation on the storage side.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

riscv,entry: fix misaligned base for excp_vect_table

[ Upstream commit ac8d0b901f0033b783156ab2dc1a0e73ec42409b ]

In RV64, the size of each entry in excp_vect_table is 8 bytes. If the
base of the table is not 8-byte aligned, loading an entry in the table
will raise a misaligned exception. Although such exception will be
handled by opensbi/bbl, this still causes performance degradation.

Signed-off-by: Zihao Yu <yuzihao@ict.ac.cn>
Reviewed-by: Anup Patel <anup@brainfault.org>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

io_uring: don't mark S_ISBLK async work as unbounded

[ Upstream commit 4b982bd0f383db9132e892c0c5144117359a6289 ]

S_ISBLK is marked as unbounded work for async preparation, because it
doesn't match S_ISREG. That is incorrect, as any read/write to a block
device is also a bounded operation. Fix it up and ensure that S_ISBLK
isn't marked unbounded.

Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

null_blk: fix command timeout completion handling

[ Upstream commit de3510e52b0a398261271455562458003b8eea62 ]

Memory backed or zoned null block devices may generate actual request
timeout errors due to the submission path being blocked on memory
allocation or zone locking. Unlike fake timeouts or injected timeouts,
the request submission path will call blk_mq_complete_request() or
blk_mq_end_request() for these real timeout errors, causing a double
completion and use after free situation as the block layer timeout
handler executes blk_mq_rq_timed_out() and __blk_mq_free_request() in
blk_mq_check_expired(). This problem often triggers a NULL pointer
dereference such as:

BUG: kernel NULL pointer dereference, address: 0000000000000050
RIP: 0010:blk_mq_sched_mark_restart_hctx+0x5/0x20
...
Call Trace:
  dd_finish_request+0x56/0x80
  blk_mq_free_request+0x37/0x130
  null_handle_cmd+0xbf/0x250 [null_blk]
  ? null_queue_rq+0x67/0xd0 [null_blk]
  blk_mq_dispatch_rq_list+0x122/0x850
  __blk_mq_do_dispatch_sched+0xbb/0x2c0
  __blk_mq_sched_dispatch_requests+0x13d/0x190
  blk_mq_sched_dispatch_requests+0x30/0x60
  __blk_mq_run_hw_queue+0x49/0x90
  process_one_work+0x26c/0x580
  worker_thread+0x55/0x3c0
  ? process_one_work+0x580/0x580
  kthread+0x134/0x150
  ? kthread_create_worker_on_cpu+0x70/0x70
  ret_from_fork+0x1f/0x30

This problem very often triggers when running the full btrfs xfstests
on a memory-backed zoned null block device in a VM with limited amount
of memory.

Avoid this by executing blk_mq_complete_request() in null_timeout_rq()
only for commands that are marked for a fake timeout completion using
the fake_timeout boolean in struct null_cmd. For timeout errors injected
through debugfs, the timeout handler will execute
blk_mq_complete_request()i as before. This is safe as the submission
path does not execute complete requests in this case.

In null_timeout_rq(), also make sure to set the command error field to
BLK_STS_TIMEOUT and to propagate this error through to the request
completion.

Reported-by: Johannes Thumshirn <Johannes.Thumshirn@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@wdc.com>
Tested-by: Johannes Thumshirn <Johannes.Thumshirn@wdc.com>
Reviewed-by: Johannes Thumshirn <Johannes.Thumshirn@wdc.com>
Link: https://lore.kernel.org/r/20210331225244.126426-1-damien.lemoal@wdc.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

idr test suite: Create anchor before launching throbber

[ Upstream commit 094ffbd1d8eaa27ed426feb8530cb1456348b018 ]

The throbber could race with creation of the anchor entry and cause the
IDR to have zero entries in it, which would cause the test to fail.

Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

idr test suite: Take RCU read lock in idr_find_test_1

[ Upstream commit 703586410da69eb40062e64d413ca33bd735917a ]

When run on a single CPU, this test would frequently access already-freed
memory.  Due to timing, this bug never showed up on multi-CPU tests.

Reported-by: Chris von Recklinghausen <crecklin@redhat.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

radix tree test suite: Register the main thread with the RCU library

[ Upstream commit 1bb4bd266cf39fd2fa711f2d265c558b92df1119 ]

Several test runners register individual worker threads with the
RCU library, but neglect to register the main thread, which can lead
to objects being freed while the main thread is in what appears to be
an RCU critical section.

Reported-by: Chris von Recklinghausen <crecklin@redhat.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

block: only update parent bi_status when bio fail

[ Upstream commit 3edf5346e4f2ce2fa0c94651a90a8dda169565ee ]

For multiple split bios, if one of the bio is fail, the whole
should return error to application. But we found there is a race
between bio_integrity_verify_fn and bio complete, which return
io success to application after one of the bio fail. The race as
following:

split bio(READ)          kworker

nvme_complete_rq
blk_update_request //split error=0
  bio_endio
    bio_integrity_endio
      queue_work(kintegrityd_wq, &bip->bip_work);

                         bio_integrity_verify_fn
                         bio_endio //split bio
                          __bio_chain_endio
                             if (!parent->bi_status)

                               <interrupt entry>
                               nvme_irq
                                 blk_update_request //parent error=7
                                 req_bio_endio
                                    bio->bi_status = 7 //parent bio
                               <interrupt exit>

                               parent->bi_status = 0
                        parent->bi_end_io() // return bi_status=0

The bio has been split as two: split and parent. When split
bio completed, it depends on kworker to do endio, while
bio_integrity_verify_fn have been interrupted by parent bio
complete irq handler. Then, parent bio->bi_status which have
been set in irq handler will overwrite by kworker.

In fact, even without the above race, we also need to conside
the concurrency beteen mulitple split bio complete and update
the same parent bi_status. Normally, multiple split bios will
be issued to the same hctx and complete from the same irq
vector. But if we have updated queue map between multiple split
bios, these bios may complete on different hw queue and different
irq vector. Then the concurrency update parent bi_status may
cause the final status error.

Suggested-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Yufen Yu <yuyufen@huawei.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Link: https://lore.kernel.org/r/20210331115359.1125679-1-yuyufen@huawei.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

XArray: Fix splitting to non-zero orders

[ Upstream commit 3012110d71f41410932924e1d188f9eb57f1f824 ]

Splitting an order-4 entry into order-2 entries would leave the array
containing pointers to 000040008000c000 instead of 000044448888cccc.
This is a one-character fix, but enhance the test suite to check this
case.

Reported-by: Zi Yan <ziy@nvidia.com>
Signed-off-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

gpu: host1x: Use different lock classes for each client

[ Upstream commit a24f98176d1efae2c37d3438c57a624d530d9c33 ]

To avoid false lockdep warnings, give each client lock a different
lock class, passed from the initialization site by macro.

Signed-off-by: Mikko Perttunen <mperttunen@nvidia.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/tegra: dc: Don't set PLL clock to 0Hz

[ Upstream commit f8fb97c915954fc6de6513cdf277103b5c6df7b3 ]

RGB output doesn't allow to change parent clock rate of the display and
PCLK rate is set to 0Hz in this case. The tegra_dc_commit_state() shall
not set the display clock to 0Hz since this change propagates to the
parent clock. The DISP clock is defined as a NODIV clock by the tegra-clk
driver and all NODIV clocks use the CLK_SET_RATE_PARENT flag.

This bug stayed unnoticed because by default PLLP is used as the parent
clock for the display controller and PLLP silently skips the erroneous 0Hz
rate changes because it always has active child clocks that don't permit
rate changes. The PLLP isn't acceptable for some devices that we want to
upstream (like Samsung Galaxy Tab and ASUS TF700T) due to a display panel
clock rate requirements that can't be fulfilled by using PLLP and then the
bug pops up in this case since parent clock is set to 0Hz, killing the
display output.

Don't touch DC clock if pclk=0 in order to fix the problem.

Signed-off-by: Dmitry Osipenko <digetx@gmail.com>
Signed-off-by: Thierry Reding <treding@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tools/kvm_stat: Add restart delay

[ Upstream commit 75f94ecbd0dfd2ac4e671f165f5ae864b7301422 ]

If this service is enabled and the system rebooted, Systemd's initial
attempt to start this unit file may fail in case the kvm module is not
loaded. Since we did not specify a delay for the retries, Systemd
restarts with a minimum delay a number of times before giving up and
disabling the service. Which means a subsequent kvm module load will
have kvm running without monitoring.
Adding a delay to fix this.

Signed-off-by: Stefan Raspl <raspl@linux.ibm.com>
Message-Id: <20210325122949.1433271-1-raspl@linux.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ftrace: Check if pages were allocated before calling free_pages()

[ Upstream commit 59300b36f85f254260c81d9dd09195fa49eb0f98 ]

It is possible that on error pg->size can be zero when getting its order,
which would return a -1 value. It is dangerous to pass in an order of -1
to free_pages(). Check if order is greater than or equal to zero before
calling free_pages().

Link: https://lore.kernel.org/lkml/20210330093916.432697c7@gandalf.local.home/
Reported-by: Abaci Robot <abaci@linux.alibaba.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

gfs2: report "already frozen/thawed" errors

[ Upstream commit ff132c5f93c06bd4432bbab5c369e468653bdec4 ]

Before this patch, gfs2's freeze function failed to report an error
when the target file system was already frozen as it should (and as
generic vfs function freeze_super does. Similarly, gfs2's thaw function
failed to report an error when trying to thaw a file system that is not
frozen, as vfs function thaw_super does. The errors were checked, but
it always returned a 0 return code.

This patch adds the missing error return codes to gfs2 freeze and thaw.

Signed-off-by: Bob Peterson <rpeterso@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/imx: imx-ldb: fix out of bounds array access warning

[ Upstream commit 33ce7f2f95cabb5834cf0906308a5cb6103976da ]

When CONFIG_OF is disabled, building with 'make W=1' produces warnings
about out of bounds array access:

drivers/gpu/drm/imx/imx-ldb.c: In function 'imx_ldb_set_clock.constprop':
drivers/gpu/drm/imx/imx-ldb.c:186:8: error: array subscript -22 is below array bounds of 'struct clk *[4]' [-Werror=array-bounds]

Add an error check before the index is used, which helps with the
warning, as well as any possible other error condition that may be
triggered at runtime.

The warning could be fixed by adding a Kconfig depedency on CONFIG_OF,
but Liu Ying points out that the driver may hit the out-of-bounds
problem at runtime anyway.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: Liu Ying <victor.liu@nxp.com>
Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: arm64: Disable guest access to trace filter controls

[ Upstream commit a354a64d91eec3e0f8ef0eed575b480fd75b999c ]

Disable guest access to the Trace Filter control registers.
We do not advertise the Trace filter feature to the guest
(ID_AA64DFR0_EL1: TRACE_FILT is cleared) already, but the guest
can still access the TRFCR_EL1 unless we trap it.

This will also make sure that the guest cannot fiddle with
the filtering controls set by a nvhe host.

Cc: Marc Zyngier <maz@kernel.org>
Cc: Will Deacon <will@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20210323120647.454211-3-suzuki.poulose@arm.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: arm64: Hide system instruction access to Trace registers

[ Upstream commit 1d676673d665fd2162e7e466dcfbe5373bfdb73e ]

Currently we advertise the ID_AA6DFR0_EL1.TRACEVER for the guest,
when the trace register accesses are trapped (CPTR_EL2.TTA == 1).
So, the guest will get an undefined instruction, if trusts the
ID registers and access one of the trace registers.
Lets be nice to the guest and hide the feature to avoid
unexpected behavior.

Even though this can be done at KVM sysreg emulation layer,
we do this by removing the TRACEVER from the sanitised feature
register field. This is fine as long as the ETM drivers
can handle the individual trace units separately, even
when there are differences among the CPUs.

Cc: Will Deacon <will@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20210323120647.454211-2-suzuki.poulose@arm.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

gfs2: Flag a withdraw if init_threads() fails

[ Upstream commit 62dd0f98a0e5668424270b47a0c2e973795faba7 ]

Interrupting mount with ^C quickly enough can cause the kthread_run()
calls in gfs2's init_threads() to fail and the error path leads to a
deadlock on the s_umount rwsem. The abridged chain of events is:

  [mount path]
  get_tree_bdev()
    sget_fc()
      alloc_super()
        down_write_nested(&s->s_umount, SINGLE_DEPTH_NESTING); [acquired]
    gfs2_fill_super()
      gfs2_make_fs_rw()
        init_threads()
          kthread_run()
            ( Interrupted )
      [Error path]
      gfs2_gl_hash_clear()
        flush_workqueue(glock_workqueue)
          wait_for_completion()

  [workqueue context]
  glock_work_func()
    run_queue()
      do_xmote()
        freeze_go_sync()
          freeze_super()
            down_write(&sb->s_umount) [deadlock]

In freeze_go_sync() there is a gfs2_withdrawn() check that we can use to
make sure freeze_super() is not called in the error path, so add a
gfs2_withdraw_delayed() call when init_threads() fails.

Ref: https://bugzilla.kernel.org/show_bug.cgi?id=212231

Reported-by: Alexander Aring <aahringo@redhat.com>
Signed-off-by: Andrew Price <anprice@redhat.com>
Signed-off-by: Andreas Gruenbacher <agruenba@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

interconnect: core: fix error return code of icc_link_destroy()

[ Upstream commit 715ea61532e731c62392221238906704e63d75b6 ]

When krealloc() fails and new is NULL, no error return code of
icc_link_destroy() is assigned.
To fix this bug, ret is assigned with -ENOMEM hen new is NULL.

Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Link: https://lore.kernel.org/r/20210306132857.17020-1-baijiaju1990@gmail.com
Signed-off-by: Georgi Djakov <georgi.djakov@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

vc4/kms: vc4_plane: Support 2020 colourspace for yuv planes

https://gist.github.com/popcornmix/6b3e23103c60170b02b148e0ba5d6ed7

is the script used to generate the 601, 709 and 2020 colourspaces.
I've regenetated the existing ones using script so it is reprocable
but there are lsb dfferences compared to values here (copied from spec)
whose origin is now lost.

Signed-off-by: Dom Cobley <popcornmix@gmail.com>

Linux 5.10.30

Tested-by: Andrei Rabusov <a.rabusov@tum.de>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Tested-by: Patrick McCormick <pmccormick@digitalocean.com>
Tested-by: Jason Self <jason@bluehome.net>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Hulk Robot <hulkrobot@huawei.com>
Tested-by: Pavel Machek (CIP) <pavel@denx.de>
Tested-by: Salvatore Bonaccorso <carnil@debian.org>
Tested-by: Sudip Mukherjee <sudip.mukherjee@codethink.co.uk>
Link: https://lore.kernel.org/r/20210412084013.643370347@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "net: sched: bump refcount for new action in ACT replace mode"

commit 4ba86128ba077fbb7d86516ae24ed642e6c3adef upstream.

This reverts commit 6855e8213e06efcaf7c02a15e12b1ae64b9a7149.

Following commit in series fixes the issue without introducing regression
in error rollback of tcf_action_destroy().

Signed-off-by: Vlad Buslov <vladbu@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ieee802154: stop dump llsec params for monitors

commit 1534efc7bbc1121e92c86c2dabebaf2c9dcece19 upstream.

This patch stops dumping llsec params for monitors which we don't support
yet. Otherwise we will access llsec mib which isn't initialized for
monitors.

Reported-by: syzbot+cde43a581a8e5f317bc2@syzkaller.appspotmail.com
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210405003054.256017-16-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ieee802154: forbid monitor for del llsec seclevel

commit 9dde130937e95b72adfae64ab21d6e7e707e2dac upstream.

This patch forbids to del llsec seclevel for monitor interfaces which we
don't support yet. Otherwise we will access llsec mib which isn't
initialized for monitors.

Reported-by: syzbot+fbf4fc11a819824e027b@syzkaller.appspotmail.com
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210405003054.256017-15-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ieee802154: forbid monitor for set llsec params

commit 88c17855ac4291fb462e13a86b7516773b6c932e upstream.

This patch forbids to set llsec params for monitor interfaces which we
don't support yet.

Reported-by: syzbot+8b6719da8a04beeafcc3@syzkaller.appspotmail.com
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210405003054.256017-3-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ieee802154: fix nl802154 del llsec devkey

commit 27c746869e1a135dffc2f2a80715bb7aa00445b4 upstream.

This patch fixes a nullpointer dereference if NL802154_ATTR_SEC_DEVKEY is
not set by the user. If this is the case nl802154 will return -EINVAL.

Reported-by: syzbot+368672e0da240db53b5f@syzkaller.appspotmail.com
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210221174321.14210-4-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ieee802154: fix nl802154 add llsec key

commit 20d5fe2d7103f5c43ad11a3d6d259e9d61165c35 upstream.

This patch fixes a nullpointer dereference if NL802154_ATTR_SEC_KEY is
not set by the user. If this is the case nl802154 will return -EINVAL.

Reported-by: syzbot+ce4e062c2d51977ddc50@syzkaller.appspotmail.com
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210221174321.14210-3-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ieee802154: fix nl802154 del llsec dev

commit 3d1eac2f45585690d942cf47fd7fbd04093ebd1b upstream.

This patch fixes a nullpointer dereference if NL802154_ATTR_SEC_DEVICE is
not set by the user. If this is the case nl802154 will return -EINVAL.

Reported-by: syzbot+d946223c2e751d136c94@syzkaller.appspotmail.com
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210221174321.14210-2-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ieee802154: fix nl802154 del llsec key

commit 37feaaf5ceb2245e474369312bb7b922ce7bce69 upstream.

This patch fixes a nullpointer dereference if NL802154_ATTR_SEC_KEY is
not set by the user. If this is the case nl802154 will return -EINVAL.

Reported-by: syzbot+ac5c11d2959a8b3c4806@syzkaller.appspotmail.com
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210221174321.14210-1-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: ieee802154: nl-mac: fix check on panid

commit 6f7f657f24405f426212c09260bf7fe8a52cef33 upstream.

This patch fixes a null pointer derefence for panid handle by move the
check for the netlink variable directly before accessing them.

Reported-by: syzbot+d4c07de0144f6f63be3a@syzkaller.appspotmail.com
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210228151817.95700-4-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: mac802154: Fix general protection fault

commit 1165affd484889d4986cf3b724318935a0b120d8 upstream.

syzbot found general protection fault in crypto_destroy_tfm()[1].
It was caused by wrong clean up loop in llsec_key_alloc().
If one of the tfm array members is in IS_ERR() range it will
cause general protection fault in clean up function [1].

Call Trace:
 crypto_free_aead include/crypto/aead.h:191 [inline] [1]
 llsec_key_alloc net/mac802154/llsec.c:156 [inline]
 mac802154_llsec_key_add+0x9e0/0xcc0 net/mac802154/llsec.c:249
 ieee802154_add_llsec_key+0x56/0x80 net/mac802154/cfg.c:338
 rdev_add_llsec_key net/ieee802154/rdev-ops.h:260 [inline]
 nl802154_add_llsec_key+0x3d3/0x560 net/ieee802154/nl802154.c:1584
 genl_family_rcv_msg_doit+0x228/0x320 net/netlink/genetlink.c:739
 genl_family_rcv_msg net/netlink/genetlink.c:783 [inline]
 genl_rcv_msg+0x328/0x580 net/netlink/genetlink.c:800
 netlink_rcv_skb+0x153/0x420 net/netlink/af_netlink.c:2502
 genl_rcv+0x24/0x40 net/netlink/genetlink.c:811
 netlink_unicast_kernel net/netlink/af_netlink.c:1312 [inline]
 netlink_unicast+0x533/0x7d0 net/netlink/af_netlink.c:1338
 netlink_sendmsg+0x856/0xd90 net/netlink/af_netlink.c:1927
 sock_sendmsg_nosec net/socket.c:654 [inline]
 sock_sendmsg+0xcf/0x120 net/socket.c:674
 ____sys_sendmsg+0x6e8/0x810 net/socket.c:2350
 ___sys_sendmsg+0xf3/0x170 net/socket.c:2404
 __sys_sendmsg+0xe5/0x1b0 net/socket.c:2433
 do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
 entry_SYSCALL_64_after_hwframe+0x44/0xae

Signed-off-by: Pavel Skripkin <paskripkin@gmail.com>
Reported-by: syzbot+9ec037722d2603a9f52e@syzkaller.appspotmail.com
Acked-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20210304152125.1052825-1-paskripkin@gmail.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers: net: fix memory leak in peak_usb_create_dev

commit a0b96b4a62745397aee662670cfc2157bac03f55 upstream.

syzbot reported memory leak in peak_usb.
The problem was in case of failure after calling
->dev_init()[2] in peak_usb_create_dev()[1]. The data
allocated int dev_init() wasn't freed, so simple
->dev_free() call fix this problem.

backtrace:
    [<0000000079d6542a>] kmalloc include/linux/slab.h:552 [inline]
    [<0000000079d6542a>] kzalloc include/linux/slab.h:682 [inline]
    [<0000000079d6542a>] pcan_usb_fd_init+0x156/0x210 drivers/net/can/usb/peak_usb/pcan_usb_fd.c:868   [2]
    [<00000000c09f9057>] peak_usb_create_dev drivers/net/can/usb/peak_usb/pcan_usb_core.c:851 [inline] [1]
    [<00000000c09f9057>] peak_usb_probe+0x389/0x490 drivers/net/can/usb/peak_usb/pcan_usb_core.c:949

Reported-by: syzbot+91adee8d9ebb9193d22d@syzkaller.appspotmail.com
Signed-off-by: Pavel Skripkin <paskripkin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drivers: net: fix memory leak in atusb_probe

commit 6b9fbe16955152626557ec6f439f3407b7769941 upstream.

syzbot reported memory leak in atusb_probe()[1].
The problem was in atusb_alloc_urbs().
Since urb is anchored, we need to release the reference
to correctly free the urb

backtrace:
    [<ffffffff82ba0466>] kmalloc include/linux/slab.h:559 [inline]
    [<ffffffff82ba0466>] usb_alloc_urb+0x66/0xe0 drivers/usb/core/urb.c:74
    [<ffffffff82ad3888>] atusb_alloc_urbs drivers/net/ieee802154/atusb.c:362 [inline][2]
    [<ffffffff82ad3888>] atusb_probe+0x158/0x820 drivers/net/ieee802154/atusb.c:1038 [1]

Reported-by: syzbot+28a246747e0a465127f3@syzkaller.appspotmail.com
Signed-off-by: Pavel Skripkin <paskripkin@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: tun: set tun->dev->addr_len during TUNSETLINK processing

commit cca8ea3b05c972ffb5295367e6c544369b45fbdd upstream.

When changing type with TUNSETLINK ioctl command, set tun->dev->addr_len
to match the appropriate type, using new tun_get_addr_len utility function
which returns appropriate address length for given type. Fixes a
KMSAN-found uninit-value bug reported by syzbot at:
https://syzkaller.appspot.com/bug?id=0766d38c656abeace60621896d705743aeefed51

Reported-by: syzbot+001516d86dbe88862cec@syzkaller.appspotmail.com
Diagnosed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Phillip Potter <phil@philpotter.co.uk>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cfg80211: remove WARN_ON() in cfg80211_sme_connect

commit 1b5ab825d9acc0f27d2f25c6252f3526832a9626 upstream.

A WARN_ON(wdev->conn) would trigger in cfg80211_sme_connect(), if multiple
send_msg(NL80211_CMD_CONNECT) system calls are made from the userland, which
should be anticipated and handled by the wireless driver. Remove this WARN_ON()
to prevent kernel panic if kernel is configured to "panic_on_warn".

Bug reported by syzbot.

Reported-by: syzbot+5f9392825de654244975@syzkaller.appspotmail.com
Signed-off-by: Du Cheng <ducheng2@gmail.com>
Link: https://lore.kernel.org/r/20210407162756.6101-1-ducheng2@gmail.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

gpiolib: Read "gpio-line-names" from a firmware node

commit b41ba2ec54a70908067034f139aa23d0dd2985ce upstream.

On STM32MP1, the GPIO banks are subnodes of pin-controller@50002000,
see arch/arm/boot/dts/stm32mp151.dtsi. The driver for
pin-controller@50002000 is in drivers/pinctrl/stm32/pinctrl-stm32.c
and iterates over all of its DT subnodes when registering each GPIO
bank gpiochip. Each gpiochip has:

  - gpio_chip.parent = dev,
    where dev is the device node of the pin controller
  - gpio_chip.of_node = np,
    which is the OF node of the GPIO bank

Therefore, dev_fwnode(chip->parent) != of_fwnode_handle(chip.of_node),
i.e. pin-controller@50002000 != pin-controller@50002000/gpio@5000*000.

The original code behaved correctly, as it extracted the "gpio-line-names"
from of_fwnode_handle(chip.of_node) = pin-controller@50002000/gpio@5000*000.

To achieve the same behaviour, read property from the firmware node.

Fixes: 7cba1a4d5e162 ("gpiolib: generalize devprop_gpiochip_set_names() for device properties")
Reported-by: Marek Vasut <marex@denx.de>
Reported-by: Roman Guskov <rguskov@dh-electronics.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Tested-by: Marek Vasut <marex@denx.de>
Reviewed-by: Marek Vasut <marex@denx.de>
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

net: sched: bump refcount for new action in ACT replace mode

commit 6855e8213e06efcaf7c02a15e12b1ae64b9a7149 upstream.

Currently, action creation using ACT API in replace mode is buggy.
When invoking for non-existent action index 42,

tc action replace action bpf obj foo.o sec <xyz> index 42

kernel creates the action, fills up the netlink response, and then just
deletes the action after notifying userspace.

tc action show action bpf

doesn't list the action.

This happens due to the following sequence when ovr = 1 (replace mode)
is enabled:

tcf_idr_check_alloc is used to atomically check and either obtain
reference for existing action at index, or reserve the index slot using
a dummy entry (ERR_PTR(-EBUSY)).

This is necessary as pointers to these actions will be held after
dropping the idrinfo lock, so bumping the reference count is necessary
as we need to insert the actions, and notify userspace by dumping their
attributes. Finally, we drop the reference we took using the
tcf_action_put_many call in tcf_action_add. However, for the case where
a new action is created due to free index, its refcount remains one.
This when paired with the put_many call leads to the kernel setting up
the action, notifying userspace of its creation, and then tearing it
down. For existing actions, the refcount is still held so they remain
unaffected.

Fortunately due to rtnl_lock serialization requirement, such an action
with refcount == 1 will not be concurrently deleted by anything else, at
best CLS API can move its refcount up and down by binding to it after it
has been published from tcf_idr_insert_many. Since refcount is atleast
one until put_many call, CLS API cannot delete it. Also __tcf_action_put
release path already ensures deterministic outcome (either new action
will be created or existing action will be reused in case CLS API tries
to bind to action concurrently) due to idr lock serialization.

We fix this by making refcount of newly created actions as 2 in ACT API
replace mode. A relaxed store will suffice as visibility is ensured only
after the tcf_idr_insert_many call.

Note that in case of creation or overwriting using CLS API only (i.e.
bind = 1), overwriting existing action object is not allowed, and any
such request is silently ignored (without error).

The refcount bump that occurs in tcf_idr_check_alloc call there for
existing action will pair with tcf_exts_destroy call made from the
owner module for the same action. In case of action creation, there
is no existing action, so no tcf_exts_destroy callback happens.

This means no code changes for CLS API.

Fixes: cae422f379f3 ("net: sched: use reference counting action init")
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

dt-bindings: net: ethernet-controller: fix typo in NVMEM

commit af9d316f3dd6d1385fbd1631b5103e620fc4298a upstream.

The correct property name is "nvmem-cell-names". This is what:
1. Was originally documented in the ethernet.txt
2. Is used in DTS files
3. Matches standard syntax for phandles
4. Linux net subsystem checks for

Fixes: 9d3de3c58347 ("dt-bindings: net: Add YAML schemas for the generic Ethernet options")
Signed-off-by: Rafał Miłecki <rafal@milecki.pl>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

lockdep: Address clang -Wformat warning printing for %hd

commit 6d48b7912cc72275dc7c59ff961c8bac7ef66a92 upstream.

Clang doesn't like format strings that truncate a 32-bit
value to something shorter:

  kernel/locking/lockdep.c:709:4: error: format specifies type 'short' but the argument has type 'int' [-Werror,-Wformat]

In this case, the warning is a slightly questionable, as it could realize
that both class->wait_type_outer and class->wait_type_inner are in fact
8-bit struct members, even though the result of the ?: operator becomes an
'int'.

However, there is really no point in printing the number as a 16-bit
'short' rather than either an 8-bit or 32-bit number, so just change
it to a normal %d.

Fixes: de8f5e4f2dc1 ("lockdep: Introduce wait-type checks")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20210322115531.3987555-1-arnd@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

clk: socfpga: fix iomem pointer cast on 64-bit

commit 2867b9746cef78745c594894aece6f8ef826e0b4 upstream.

Pointers should be cast with uintptr_t instead of integer.  This fixes
warning when compile testing on ARM64:

  drivers/clk/socfpga/clk-gate.c: In function ‘socfpga_clk_recalc_rate’:
  drivers/clk/socfpga/clk-gate.c:102:7: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast]

Fixes: b7cec13f082f ("clk: socfpga: Look for the GPIO_DB_CLK by its offset")
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@canonical.com>
Acked-by: Dinh Nguyen <dinguyen@kernel.org>
Link: https://lore.kernel.org/r/20210314110709.32599-1-krzysztof.kozlowski@canonical.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

RAS/CEC: Correct ce_add_elem()'s returned values

commit 3a62583c2853b0ab37a57dde79decea210b5fb89 upstream.

ce_add_elem() uses different return values to signal a result from
adding an element to the collector. Commit in Fixes: broke the case
where the element being added is not found in the array. Correct that.

 [ bp: Rewrite commit message, add kernel-doc comments. ]

Fixes: de0e0624d86f ("RAS/CEC: Check count_threshold unconditionally")
Signed-off-by: William Roche <william.roche@oracle.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: <stable@vger.kernel.org>
Link: https://lkml.kernel.org/r/1617722939-29670-1-git-send-email-william.roche@oracle.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

vdpa/mlx5: Fix wrong use of bit numbers

[ Upstream commit 4b454a82418dd76d8c0590bb3f7a99a63ea57dc5 ]

VIRTIO_F_VERSION_1 is a bit number. Use BIT_ULL() with mask
conditionals.

Also, in mlx5_vdpa_is_little_endian() use BIT_ULL for consistency with
the rest of the code.

Fixes: 1a86b377aa21 ("vdpa/mlx5: Add VDPA driver for supported mlx5 devices")
Signed-off-by: Eli Cohen <elic@nvidia.com>
Link: https://lore.kernel.org/r/20210408091047.4269-5-elic@nvidia.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

vdpa/mlx5: should exclude header length and fcs from mtu

[ Upstream commit d084d996aaf53c0cc583dc75a4fc2a67fe485846 ]

When feature VIRTIO_NET_F_MTU is negotiated on mlx5_vdpa,
22 extra bytes worth of MTU length is shown in guest.
This is because the mlx5_query_port_max_mtu API returns
the "hardware" MTU value, which does not just contain the
 Ethernet payload, but includes extra lengths starting
from the Ethernet header up to the FCS altogether.

Fix the MTU so packets won't get dropped silently.

Fixes: 1a86b377aa21 ("vdpa/mlx5: Add VDPA driver for supported mlx5 devices")
Signed-off-by: Si-Wei Liu <si-wei.liu@oracle.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Acked-by: Eli Cohen <elic@nvidia.com>
Link: https://lore.kernel.org/r/20210408091047.4269-2-elic@nvidia.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/addr: Be strict with gid size

[ Upstream commit d1c803a9ccd7bd3aff5e989ccfb39ed3b799b975 ]

The nla_len() is less than or equal to 16.  If it's less than 16 then end
of the "gid" buffer is uninitialized.

Fixes: ae43f8286730 ("IB/core: Add IP to GID netlink offload")
Link: https://lore.kernel.org/r/20210405074434.264221-1-leon@kernel.org
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Mark Bloch <mbloch@nvidia.com>
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i40e: Fix parameters in aq_get_phy_register()

[ Upstream commit b2d0efc4be7ed320e33eaa9b6dd6f3f6011ffb8e ]

Change parameters order in aq_get_phy_register() due to wrong
statistics in PHY reported by ethtool. Previously all PHY statistics were
exactly the same for all interfaces
Now statistics are reported correctly - different for different interfaces

Fixes: 0514db37dd78 ("i40e: Extend PHY access with page change flag")
Signed-off-by: Grzegorz Siwik <grzegorz.siwik@intel.com>
Tested-by: Dave Switzer <david.switzer@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/vc4: crtc: Reduce PV fifo threshold on hvs4

[ Upstream commit eb9dfdd1ed40357b99a4201c8534c58c562e48c9 ]

Experimentally have found PV on hvs4 reports fifo full
error with expected settings and does not with one less

This appears as:
[drm:drm_atomic_helper_wait_for_flip_done] *ERROR* [CRTC:82:crtc-3] flip_done timed out

with bit 10 of PV_STAT set "HVS driving pixels when the PV FIFO is full"

Fixes: c8b75bca92cb ("drm/vc4: Add KMS support for Raspberry Pi.")
Signed-off-by: Dom Cobley <popcornmix@gmail.com>
Signed-off-by: Maxime Ripard <maxime@cerno.tech>
Link: https://patchwork.freedesktop.org/patch/msgid/20210318161328.1471556-3-maxime@cerno.tech
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/qedr: Fix kernel panic when trying to access recv_cq

[ Upstream commit e1ad897b9c738d5550be6762bf3a6ef1672259a4 ]

As INI QP does not require a recv_cq, avoid the following null pointer
dereference by checking if the qp_type is not INI before trying to extract
the recv_cq.

BUG: kernel NULL pointer dereference, address: 00000000000000e0
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0
 Oops: 0000 [#1] SMP PTI
 CPU: 0 PID: 54250 Comm: mpitests-IMB-MP Not tainted 5.12.0-rc5 #1
 Hardware name: Dell Inc. PowerEdge R320/0KM5PX, BIOS 2.7.0 08/19/2019
 RIP: 0010:qedr_create_qp+0x378/0x820 [qedr]
 Code: 02 00 00 50 e8 29 d4 a9 d1 48 83 c4 18 e9 65 fe ff ff 48 8b 53 10 48 8b 43 18 44 8b 82 e0 00 00 00 45 85 c0 0f 84 10 74 00 00 <8b> b8 e0 00 00 00 85 ff 0f 85 50 fd ff ff e9 fd 73 00 00 48 8d bd
 RSP: 0018:ffff9c8f056f7a70 EFLAGS: 00010202
 RAX: 0000000000000000 RBX: ffff9c8f056f7b58 RCX: 0000000000000009
 RDX: ffff8c41a9744c00 RSI: ffff9c8f056f7b58 RDI: ffff8c41c0dfa280
 RBP: ffff8c41c0dfa280 R08: 0000000000000002 R09: 0000000000000001
 R10: 0000000000000000 R11: ffff8c41e06fc608 R12: ffff8c4194052000
 R13: 0000000000000000 R14: ffff8c4191546070 R15: ffff8c41c0dfa280
 FS:  00007f78b2787b80(0000) GS:ffff8c43a3200000(0000) knlGS:0000000000000000
 CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 CR2: 00000000000000e0 CR3: 00000001011d6002 CR4: 00000000001706f0
 Call Trace:
  ib_uverbs_handler_UVERBS_METHOD_QP_CREATE+0x4e4/0xb90 [ib_uverbs]
  ? ib_uverbs_cq_event_handler+0x30/0x30 [ib_uverbs]
  ib_uverbs_run_method+0x6f6/0x7a0 [ib_uverbs]
  ? ib_uverbs_handler_UVERBS_METHOD_QP_DESTROY+0x70/0x70 [ib_uverbs]
  ? __cond_resched+0x15/0x30
  ? __kmalloc+0x5a/0x440
  ib_uverbs_cmd_verbs+0x195/0x360 [ib_uverbs]
  ? xa_load+0x6e/0x90
  ? cred_has_capability+0x7c/0x130
  ? avc_has_extended_perms+0x17f/0x440
  ? vma_link+0xae/0xb0
  ? vma_set_page_prot+0x2a/0x60
  ? mmap_region+0x298/0x6c0
  ? do_mmap+0x373/0x520
  ? selinux_file_ioctl+0x17f/0x220
  ib_uverbs_ioctl+0xa7/0x110 [ib_uverbs]
  __x64_sys_ioctl+0x84/0xc0
  do_syscall_64+0x33/0x40
  entry_SYSCALL_64_after_hwframe+0x44/0xae
 RIP: 0033:0x7f78b120262b

Fixes: 06e8d1df46ed ("RDMA/qedr: Add support for user mode XRC-SRQ's")
Link: https://lore.kernel.org/r/20210404125501.154789-1-kamalheib1@gmail.com
Signed-off-by: Kamal Heib <kamalheib1@gmail.com>
Signed-off-by: Jason Gunthorpe <jgg@nvidia.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf report: Fix wrong LBR block sorting

[ Upstream commit f2013278ae40b89cc27916366c407ce5261815ef ]

When '--total-cycles' is specified, it supports sorting for all blocks
by 'Sampled Cycles%'. This is useful to concentrate on the globally
hottest blocks.

'Sampled Cycles%' - block sampled cycles aggregation / total sampled cycles

But in current code, it doesn't use the cycles aggregation. Part of
'cycles' counting is possibly dropped for some overlap jumps. But for
identifying the hot block, we always need the full cycles.

  # perf record -b ./triad_loop
  # perf report --total-cycles --stdio

Before:

  #
  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                                          [Program Block Range]      Shared Object
  # ...............  ..............  ...........  ..........  .............................................................  .................
  #
              0.81%             793        4.32%         793                           [setup-vdso.h:34 -> setup-vdso.h:40]         ld-2.27.so
              0.49%             480        0.87%         160                    [native_write_msr+0 -> native_write_msr+16]  [kernel.kallsyms]
              0.48%             476        0.52%          95                      [native_read_msr+0 -> native_read_msr+29]  [kernel.kallsyms]
              0.31%             303        1.65%         303                              [nmi_restore+0 -> nmi_restore+37]  [kernel.kallsyms]
              0.26%             255        1.39%         255      [nohz_balance_exit_idle+75 -> nohz_balance_exit_idle+162]  [kernel.kallsyms]
              0.24%             234        1.28%         234                       [end_repeat_nmi+67 -> end_repeat_nmi+83]  [kernel.kallsyms]
              0.23%             227        1.24%         227            [__irqentry_text_end+96 -> __irqentry_text_end+126]  [kernel.kallsyms]
              0.20%             194        1.06%         194             [native_set_debugreg+52 -> native_set_debugreg+56]  [kernel.kallsyms]
              0.11%             106        0.14%          26                [native_sched_clock+0 -> native_sched_clock+98]  [kernel.kallsyms]
              0.10%              97        0.53%          97            [trigger_load_balance+0 -> trigger_load_balance+67]  [kernel.kallsyms]
              0.09%              85        0.46%          85             [get-dynamic-info.h:102 -> get-dynamic-info.h:111]         ld-2.27.so
  ...
              0.00%           92.7K        0.02%           4                           [triad_loop.c:64 -> triad_loop.c:65]         triad_loop

The hottest block '[triad_loop.c:64 -> triad_loop.c:65]' is not at
the top of output.

After:

  # Sampled Cycles%  Sampled Cycles  Avg Cycles%  Avg Cycles                                           [Program Block Range]      Shared Object
  # ...............  ..............  ...........  ..........  ..............................................................  .................
  #
             94.35%           92.7K        0.02%           4                            [triad_loop.c:64 -> triad_loop.c:65]         triad_loop
              0.81%             793        4.32%         793                            [setup-vdso.h:34 -> setup-vdso.h:40]         ld-2.27.so
              0.49%             480        0.87%         160                     [native_write_msr+0 -> native_write_msr+16]  [kernel.kallsyms]
              0.48%             476        0.52%          95                       [native_read_msr+0 -> native_read_msr+29]  [kernel.kallsyms]
              0.31%             303        1.65%         303                               [nmi_restore+0 -> nmi_restore+37]  [kernel.kallsyms]
              0.26%             255        1.39%         255       [nohz_balance_exit_idle+75 -> nohz_balance_exit_idle+162]  [kernel.kallsyms]
              0.24%             234        1.28%         234                        [end_repeat_nmi+67 -> end_repeat_nmi+83]  [kernel.kallsyms]
              0.23%             227        1.24%         227             [__irqentry_text_end+96 -> __irqentry_text_end+126]  [kernel.kallsyms]
              0.20%             194        1.06%         194              [native_set_debugreg+52 -> native_set_debugreg+56]  [kernel.kallsyms]
              0.11%             106        0.14%          26                 [native_sched_clock+0 -> native_sched_clock+98]  [kernel.kallsyms]
              0.10%              97        0.53%          97             [trigger_load_balance+0 -> trigger_load_balance+67]  [kernel.kallsyms]
              0.09%              85        0.46%          85              [get-dynamic-info.h:102 -> get-dynamic-info.h:111]         ld-2.27.so
              0.08%              82        0.06%          11  [intel_pmu_drain_pebs_nhm+580 -> intel_pmu_drain_pebs_nhm+627]  [kernel.kallsyms]
              0.08%              77        0.42%          77                  [lru_add_drain_cpu+0 -> lru_add_drain_cpu+133]  [kernel.kallsyms]
              0.08%              74        0.10%          18                [handle_pmi_common+271 -> handle_pmi_common+310]  [kernel.kallsyms]
              0.08%              74        0.40%          74              [get-dynamic-info.h:131 -> get-dynamic-info.h:157]         ld-2.27.so
              0.07%              69        0.09%          17  [intel_pmu_drain_pebs_nhm+432 -> intel_pmu_drain_pebs_nhm+468]  [kernel.kallsyms]

Now the hottest block is reported at the top of output.

Fixes: b65a7d372b1a55db ("perf hist: Support block formats with compare/sort/display")
Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
Reviewed-by: Andi Kleen <ak@linux.intel.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Jin Yao <yao.jin@intel.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lore.kernel.org/lkml/20210407024452.29988-1-yao.jin@linux.intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>