mbox series

[v1,0/2] amd-pstate-ut: Fix issues observed after loading amd-pstate-ut driver

Message ID 20230818114452.66780-1-swapnil.sapkal@amd.com
Headers show
Series amd-pstate-ut: Fix issues observed after loading amd-pstate-ut driver | expand

Message

Swapnil Sapkal Aug. 18, 2023, 11:44 a.m. UTC 
This series fixes amd-pstate-ut which is currently failing for two
reasons:

1) The code for amd-pstate-ut was written when CONFIG_X86_AMD_PSTATE
was tristate config and amd_pstate can be built as module. In
current implementation the driver is built-in and previous module
parameters are removed. But amd-pstate-ut still try to read old module
parameters. Therefore, test fails. First patch fixes this issue by
removing those module parameter references.

2) In amd_pstate_ut_check_perf() and amd_pstate_ut_check_freq()
functions, cpufreq_cpu_get() is called to get policy of the
cpu and mark it as busy. In these functions cpufreq_cpu_put()
should be used to release the policy. As cpufreq_cpu_put()
is not used to release the policy, any other entity trying to
access the policy is blocked indefinitely. One such scenario
occurs when mode of amd-pstate is changed. Second patch fixes this
by calling cpufreq_cpu_put() wherever necessary.

Swapnil Sapkal (2):
  amd-pstate-ut: Remove module parameter access from amd-pstate-ut
  amd-pstate-ut: Fix kernel panic due to loading amd-pstate-ut driver

 drivers/cpufreq/amd-pstate-ut.c | 46 +++++++++++++--------------------
 1 file changed, 18 insertions(+), 28 deletions(-)

Comments

Rafael J. Wysocki Aug. 22, 2023, 8:24 p.m. UTC | #1 
On Fri, Aug 18, 2023 at 1:45 PM Swapnil Sapkal <swapnil.sapkal@amd.com> wrote:
>
> This series fixes amd-pstate-ut which is currently failing for two
> reasons:
>
> 1) The code for amd-pstate-ut was written when CONFIG_X86_AMD_PSTATE
> was tristate config and amd_pstate can be built as module. In
> current implementation the driver is built-in and previous module
> parameters are removed. But amd-pstate-ut still try to read old module
> parameters. Therefore, test fails. First patch fixes this issue by
> removing those module parameter references.
>
> 2) In amd_pstate_ut_check_perf() and amd_pstate_ut_check_freq()
> functions, cpufreq_cpu_get() is called to get policy of the
> cpu and mark it as busy. In these functions cpufreq_cpu_put()
> should be used to release the policy. As cpufreq_cpu_put()
> is not used to release the policy, any other entity trying to
> access the policy is blocked indefinitely. One such scenario
> occurs when mode of amd-pstate is changed. Second patch fixes this
> by calling cpufreq_cpu_put() wherever necessary.
>
> Swapnil Sapkal (2):
>   amd-pstate-ut: Remove module parameter access from amd-pstate-ut
>   amd-pstate-ut: Fix kernel panic due to loading amd-pstate-ut driver
>
>  drivers/cpufreq/amd-pstate-ut.c | 46 +++++++++++++--------------------
>  1 file changed, 18 insertions(+), 28 deletions(-)
>
> --

Both patches applied as 6.6 material with subjects adjusted to the
usual cpufreq patch naming pattern and some changelog edits.

Thanks!