s390/vdso: fix initializing and updating of vdso_data
authorHeiko Carstens <hca@linux.ibm.com>
Wed, 24 Mar 2021 19:23:55 +0000 (20:23 +0100)
committerHeiko Carstens <hca@linux.ibm.com>
Thu, 25 Mar 2021 20:57:26 +0000 (21:57 +0100)
commit5b43bd184530af6b868d8273b0a743a138d37ee8
treef8420c59764e10b88616096c178fd8875d5d2ee9
parentb24bacd67ffddd9192c4745500fd6f73dbfe565e
s390/vdso: fix initializing and updating of vdso_data

Li Wang reported that clock_gettime(CLOCK_MONOTONIC_RAW, ...) returns
incorrect values when time is provided via vdso instead of system call:

vdso_ts_nsec = 4484351380985507, vdso_ts.tv_sec = 4484351, vdso_ts.tv_nsec = 380985507
sys_ts_nsec  = 1446923235377, sys_ts.tv_sec  = 1446, sys_ts.tv_nsec  = 923235377

Within the s390 specific vdso function __arch_get_hw_counter() reads
tod clock steering values from the arch_data member of the passed in
vdso_data structure.

Problem is that only for the CS_HRES_COARSE vdso_data arch_data is
initialized and gets updated. The CS_RAW specific vdso_data does not
contain any valid tod_clock_steering information, which explains the
different values.

Fix this by initializing and updating all vdso_datas.

Reported-by: Li Wang <liwang@redhat.com>
Tested-by: Li Wang <liwang@redhat.com>
Fixes: 1ba2d6c0fd4e ("s390/vdso: simplify __arch_get_hw_counter()")
Link: https://lore.kernel.org/linux-s390/YFnxr1ZlMIOIqjfq@osiris
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
arch/s390/kernel/time.c