| Message ID | 20211208115435.610101-1-james.clark@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | perf cs-etm: Remove duplicate and incorrect aux size checks | expand |
Hi James, On Wed, Dec 08, 2021 at 11:54:35AM +0000, James Clark wrote: > There are two checks, one is for size when running without admin, but > this one is covered by the driver and reported on in more detail here > (builtin-record.c): > > pr_err("Permission error mapping pages.\n" > "Consider increasing " > "/proc/sys/kernel/perf_event_mlock_kb,\n" > "or try again with a smaller value of -m/--mmap_pages.\n" > "(current value: %u,%u)\n", I looked into the kernel code and found: sysctl_perf_event_mlock = 512 + (PAGE_SIZE / 1024); // 512KB + 1 page If the system have multiple cores, let's say 8 cores, then kernel even can relax the limitaion with: user_lock_limit *= num_online_cpus(); So means the memory lock limitation is: (512KB + 1 page) * 8 = 4MB + 8 pages. Seems to me, it's much relax than the user space's limitaion 128KB. And let's imagine for Arm server, the permitted buffer size can be a huge value (e.g. for a system with 128 cores). Could you confirm if this is right? Thanks, Leo > This had the effect of artificially limiting the aux buffer size to a > value smaller than what was allowed because perf_event_mlock_kb wasn't > taken into account. > > The second is to check for a power of two, but this is covered here > (evlist.c): > > pr_info("rounding mmap pages size to %s (%lu pages)\n", > buf, pages); > > Signed-off-by: James Clark <james.clark@arm.com> > --- > tools/perf/arch/arm/util/cs-etm.c | 19 ------------------- > 1 file changed, 19 deletions(-) > > diff --git a/tools/perf/arch/arm/util/cs-etm.c b/tools/perf/arch/arm/util/cs-etm.c > index 293a23bf8be3..8a3d54a86c9c 100644 > --- a/tools/perf/arch/arm/util/cs-etm.c > +++ b/tools/perf/arch/arm/util/cs-etm.c > @@ -407,25 +407,6 @@ static int cs_etm_recording_options(struct auxtrace_record *itr, > > } > > - /* Validate auxtrace_mmap_pages provided by user */ > - if (opts->auxtrace_mmap_pages) { > - unsigned int max_page = (KiB(128) / page_size); > - size_t sz = opts->auxtrace_mmap_pages * (size_t)page_size; > - > - if (!privileged && > - opts->auxtrace_mmap_pages > max_page) { > - opts->auxtrace_mmap_pages = max_page; > - pr_err("auxtrace too big, truncating to %d\n", > - max_page); > - } > - > - if (!is_power_of_2(sz)) { > - pr_err("Invalid mmap size for %s: must be a power of 2\n", > - CORESIGHT_ETM_PMU_NAME); > - return -EINVAL; > - } > - } > - > if (opts->auxtrace_snapshot_mode) > pr_debug2("%s snapshot size: %zu\n", CORESIGHT_ETM_PMU_NAME, > opts->auxtrace_snapshot_size); > -- > 2.28.0 >
On 08/12/2021 13:17, Leo Yan wrote: > Hi James, > > On Wed, Dec 08, 2021 at 11:54:35AM +0000, James Clark wrote: >> There are two checks, one is for size when running without admin, but >> this one is covered by the driver and reported on in more detail here >> (builtin-record.c): >> >> pr_err("Permission error mapping pages.\n" >> "Consider increasing " >> "/proc/sys/kernel/perf_event_mlock_kb,\n" >> "or try again with a smaller value of -m/--mmap_pages.\n" >> "(current value: %u,%u)\n", > > I looked into the kernel code and found: > > sysctl_perf_event_mlock = 512 + (PAGE_SIZE / 1024); // 512KB + 1 page > > If the system have multiple cores, let's say 8 cores, then kernel even > can relax the limitaion with: > > user_lock_limit *= num_online_cpus(); > > So means the memory lock limitation is: > > (512KB + 1 page) * 8 = 4MB + 8 pages. > > Seems to me, it's much relax than the user space's limitaion 128KB. > And let's imagine for Arm server, the permitted buffer size can be a > huge value (e.g. for a system with 128 cores). > > Could you confirm if this is right? Yes that seems to be the case. And the commit message for that addition states the reasoning: perf_counter: Increase mmap limit In a default 'perf top' run the tool will create a counter for each online CPU. With enough CPUs this will eventually exhaust the default limit. So scale it up with the number of online CPUs. To me that makes sense. Normally the memory installed also scales with the number of cores. Are you saying that we should look into modifying that scaling factor in perf_mmap()? Or that we should still add something to userspace for coresight to limit user supplied buffer sizes? I think it makes sense to allow the user to specify any value that will work, it's up to them. James > > Thanks, > Leo > >> This had the effect of artificially limiting the aux buffer size to a >> value smaller than what was allowed because perf_event_mlock_kb wasn't >> taken into account. >> >> The second is to check for a power of two, but this is covered here >> (evlist.c): >> >> pr_info("rounding mmap pages size to %s (%lu pages)\n", >> buf, pages); >> >> Signed-off-by: James Clark <james.clark@arm.com> >> --- >> tools/perf/arch/arm/util/cs-etm.c | 19 ------------------- >> 1 file changed, 19 deletions(-) >> >> diff --git a/tools/perf/arch/arm/util/cs-etm.c b/tools/perf/arch/arm/util/cs-etm.c >> index 293a23bf8be3..8a3d54a86c9c 100644 >> --- a/tools/perf/arch/arm/util/cs-etm.c >> +++ b/tools/perf/arch/arm/util/cs-etm.c >> @@ -407,25 +407,6 @@ static int cs_etm_recording_options(struct auxtrace_record *itr, >> >> } >> >> - /* Validate auxtrace_mmap_pages provided by user */ >> - if (opts->auxtrace_mmap_pages) { >> - unsigned int max_page = (KiB(128) / page_size); >> - size_t sz = opts->auxtrace_mmap_pages * (size_t)page_size; >> - >> - if (!privileged && >> - opts->auxtrace_mmap_pages > max_page) { >> - opts->auxtrace_mmap_pages = max_page; >> - pr_err("auxtrace too big, truncating to %d\n", >> - max_page); >> - } >> - >> - if (!is_power_of_2(sz)) { >> - pr_err("Invalid mmap size for %s: must be a power of 2\n", >> - CORESIGHT_ETM_PMU_NAME); >> - return -EINVAL; >> - } >> - } >> - >> if (opts->auxtrace_snapshot_mode) >> pr_debug2("%s snapshot size: %zu\n", CORESIGHT_ETM_PMU_NAME, >> opts->auxtrace_snapshot_size); >> -- >> 2.28.0 >>
On Wed, Dec 08, 2021 at 02:08:04PM +0000, James Clark wrote: > On 08/12/2021 13:17, Leo Yan wrote: > > Hi James, > > > > On Wed, Dec 08, 2021 at 11:54:35AM +0000, James Clark wrote: > >> There are two checks, one is for size when running without admin, but > >> this one is covered by the driver and reported on in more detail here > >> (builtin-record.c): > >> > >> pr_err("Permission error mapping pages.\n" > >> "Consider increasing " > >> "/proc/sys/kernel/perf_event_mlock_kb,\n" > >> "or try again with a smaller value of -m/--mmap_pages.\n" > >> "(current value: %u,%u)\n", > > > > I looked into the kernel code and found: > > > > sysctl_perf_event_mlock = 512 + (PAGE_SIZE / 1024); // 512KB + 1 page > > > > If the system have multiple cores, let's say 8 cores, then kernel even > > can relax the limitaion with: > > > > user_lock_limit *= num_online_cpus(); > > > > So means the memory lock limitation is: > > > > (512KB + 1 page) * 8 = 4MB + 8 pages. > > > > Seems to me, it's much relax than the user space's limitaion 128KB. > > And let's imagine for Arm server, the permitted buffer size can be a > > huge value (e.g. for a system with 128 cores). > > > > Could you confirm if this is right? > > Yes that seems to be the case. And the commit message for that addition > states the reasoning: > > perf_counter: Increase mmap limit > > In a default 'perf top' run the tool will create a counter for > each online CPU. With enough CPUs this will eventually exhaust > the default limit. > > So scale it up with the number of online CPUs. > > To me that makes sense. Normally the memory installed also scales with the > number of cores. > > Are you saying that we should look into modifying that scaling factor in > perf_mmap()? Or that we should still add something to userspace for > coresight to limit user supplied buffer sizes? I don't think we should modify the scaling factor in perf_mmap(), the logic is not only used by AUX buffer, it's shared by normal event ring buffer. > I think it makes sense to allow the user to specify any value that will work, > it's up to them. Understand, I verified this patch with below steps: root@debian:~# echo 0 > /proc/sys/kernel/perf_event_paranoid leoy@debian:~$ perf record -e cs_etm// -m 4M,8M -o perf_test.data -- sleep 1 Permission error mapping pages. Consider increasing /proc/sys/kernel/perf_event_mlock_kb, or try again with a smaller value of -m/--mmap_pages. (current value: 1024,2048) leoy@debian:~$ perf record -e cs_etm// -m 4M,4M -o perf_test.data -- sleep 1 Couldn't synthesize bpf events. [ perf record: Woken up 1 times to write data ] [ perf record: Captured and wrote 0.607 MB perf_test.data ] So this patch looks good for me: Reviewed-by: Leo Yan <leo.yan@linaro.org>
On 09/12/2021 13:44, Leo Yan wrote: > On Wed, Dec 08, 2021 at 02:08:04PM +0000, James Clark wrote: >> On 08/12/2021 13:17, Leo Yan wrote: >>> Hi James, >>> >>> On Wed, Dec 08, 2021 at 11:54:35AM +0000, James Clark wrote: >>>> There are two checks, one is for size when running without admin, but >>>> this one is covered by the driver and reported on in more detail here >>>> (builtin-record.c): >>>> >>>> pr_err("Permission error mapping pages.\n" >>>> "Consider increasing " >>>> "/proc/sys/kernel/perf_event_mlock_kb,\n" >>>> "or try again with a smaller value of -m/--mmap_pages.\n" >>>> "(current value: %u,%u)\n", >>> >>> I looked into the kernel code and found: >>> >>> sysctl_perf_event_mlock = 512 + (PAGE_SIZE / 1024); // 512KB + 1 page >>> >>> If the system have multiple cores, let's say 8 cores, then kernel even >>> can relax the limitaion with: >>> >>> user_lock_limit *= num_online_cpus(); >>> >>> So means the memory lock limitation is: >>> >>> (512KB + 1 page) * 8 = 4MB + 8 pages. >>> >>> Seems to me, it's much relax than the user space's limitaion 128KB. >>> And let's imagine for Arm server, the permitted buffer size can be a >>> huge value (e.g. for a system with 128 cores). >>> >>> Could you confirm if this is right? >> >> Yes that seems to be the case. And the commit message for that addition >> states the reasoning: >> >> perf_counter: Increase mmap limit >> >> In a default 'perf top' run the tool will create a counter for >> each online CPU. With enough CPUs this will eventually exhaust >> the default limit. >> >> So scale it up with the number of online CPUs. >> >> To me that makes sense. Normally the memory installed also scales with the >> number of cores. >> >> Are you saying that we should look into modifying that scaling factor in >> perf_mmap()? Or that we should still add something to userspace for >> coresight to limit user supplied buffer sizes? > > I don't think we should modify the scaling factor in perf_mmap(), the > logic is not only used by AUX buffer, it's shared by normal event > ring buffer. > >> I think it makes sense to allow the user to specify any value that will work, >> it's up to them. > > Understand, I verified this patch with below steps: > > root@debian:~# echo 0 > /proc/sys/kernel/perf_event_paranoid > > leoy@debian:~$ perf record -e cs_etm// -m 4M,8M -o perf_test.data -- sleep 1 > Permission error mapping pages. > Consider increasing /proc/sys/kernel/perf_event_mlock_kb, > or try again with a smaller value of -m/--mmap_pages. > (current value: 1024,2048) > > leoy@debian:~$ perf record -e cs_etm// -m 4M,4M -o perf_test.data -- sleep 1 > Couldn't synthesize bpf events. > [ perf record: Woken up 1 times to write data ] > [ perf record: Captured and wrote 0.607 MB perf_test.data ] > > So this patch looks good for me: > > Reviewed-by: Leo Yan <leo.yan@linaro.org> > Thanks Leo!
On Thu, Dec 09, 2021 at 02:16:43PM +0000, James Clark wrote: > > > On 09/12/2021 13:44, Leo Yan wrote: > > On Wed, Dec 08, 2021 at 02:08:04PM +0000, James Clark wrote: > >> On 08/12/2021 13:17, Leo Yan wrote: > >>> Hi James, > >>> > >>> On Wed, Dec 08, 2021 at 11:54:35AM +0000, James Clark wrote: > >>>> There are two checks, one is for size when running without admin, but > >>>> this one is covered by the driver and reported on in more detail here > >>>> (builtin-record.c): > >>>> > >>>> pr_err("Permission error mapping pages.\n" > >>>> "Consider increasing " > >>>> "/proc/sys/kernel/perf_event_mlock_kb,\n" > >>>> "or try again with a smaller value of -m/--mmap_pages.\n" > >>>> "(current value: %u,%u)\n", > >>> > >>> I looked into the kernel code and found: > >>> > >>> sysctl_perf_event_mlock = 512 + (PAGE_SIZE / 1024); // 512KB + 1 page > >>> > >>> If the system have multiple cores, let's say 8 cores, then kernel even > >>> can relax the limitaion with: > >>> > >>> user_lock_limit *= num_online_cpus(); > >>> > >>> So means the memory lock limitation is: > >>> > >>> (512KB + 1 page) * 8 = 4MB + 8 pages. > >>> > >>> Seems to me, it's much relax than the user space's limitaion 128KB. > >>> And let's imagine for Arm server, the permitted buffer size can be a > >>> huge value (e.g. for a system with 128 cores). > >>> > >>> Could you confirm if this is right? > >> > >> Yes that seems to be the case. And the commit message for that addition > >> states the reasoning: > >> > >> perf_counter: Increase mmap limit > >> > >> In a default 'perf top' run the tool will create a counter for > >> each online CPU. With enough CPUs this will eventually exhaust > >> the default limit. > >> > >> So scale it up with the number of online CPUs. > >> > >> To me that makes sense. Normally the memory installed also scales with the > >> number of cores. > >> > >> Are you saying that we should look into modifying that scaling factor in > >> perf_mmap()? Or that we should still add something to userspace for > >> coresight to limit user supplied buffer sizes? > > > > I don't think we should modify the scaling factor in perf_mmap(), the > > logic is not only used by AUX buffer, it's shared by normal event > > ring buffer. > > > >> I think it makes sense to allow the user to specify any value that will work, > >> it's up to them. > > > > Understand, I verified this patch with below steps: > > > > root@debian:~# echo 0 > /proc/sys/kernel/perf_event_paranoid > > > > leoy@debian:~$ perf record -e cs_etm// -m 4M,8M -o perf_test.data -- sleep 1 > > Permission error mapping pages. > > Consider increasing /proc/sys/kernel/perf_event_mlock_kb, > > or try again with a smaller value of -m/--mmap_pages. > > (current value: 1024,2048) > > > > leoy@debian:~$ perf record -e cs_etm// -m 4M,4M -o perf_test.data -- sleep 1 > > Couldn't synthesize bpf events. > > [ perf record: Woken up 1 times to write data ] > > [ perf record: Captured and wrote 0.607 MB perf_test.data ] > > > > So this patch looks good for me: > > > > Reviewed-by: Leo Yan <leo.yan@linaro.org> > > > Thanks Leo! Arnaldo is not on the recipient list and as such he won't see this patch...
On December 10, 2021 1:54:36 PM GMT-03:00, Mathieu Poirier <mathieu.poirier@linaro.org> wrote: >On Thu, Dec 09, 2021 at 02:16:43PM +0000, James Clark wrote: >> >> >> On 09/12/2021 13:44, Leo Yan wrote: >> > On Wed, Dec 08, 2021 at 02:08:04PM +0000, James Clark wrote: >> >> On 08/12/2021 13:17, Leo Yan wrote: >> >>> Hi James, >> >>> >> >>> On Wed, Dec 08, 2021 at 11:54:35AM +0000, James Clark wrote: >> >>>> There are two checks, one is for size when running without admin, but >> >>>> this one is covered by the driver and reported on in more detail here >> >>>> (builtin-record.c): >> >>>> >> >>>> pr_err("Permission error mapping pages.\n" >> >>>> "Consider increasing " >> >>>> "/proc/sys/kernel/perf_event_mlock_kb,\n" >> >>>> "or try again with a smaller value of -m/--mmap_pages.\n" >> >>>> "(current value: %u,%u)\n", >> >>> >> >>> I looked into the kernel code and found: >> >>> >> >>> sysctl_perf_event_mlock = 512 + (PAGE_SIZE / 1024); // 512KB + 1 page >> >>> >> >>> If the system have multiple cores, let's say 8 cores, then kernel even >> >>> can relax the limitaion with: >> >>> >> >>> user_lock_limit *= num_online_cpus(); >> >>> >> >>> So means the memory lock limitation is: >> >>> >> >>> (512KB + 1 page) * 8 = 4MB + 8 pages. >> >>> >> >>> Seems to me, it's much relax than the user space's limitaion 128KB. >> >>> And let's imagine for Arm server, the permitted buffer size can be a >> >>> huge value (e.g. for a system with 128 cores). >> >>> >> >>> Could you confirm if this is right? >> >> >> >> Yes that seems to be the case. And the commit message for that addition >> >> states the reasoning: >> >> >> >> perf_counter: Increase mmap limit >> >> >> >> In a default 'perf top' run the tool will create a counter for >> >> each online CPU. With enough CPUs this will eventually exhaust >> >> the default limit. >> >> >> >> So scale it up with the number of online CPUs. >> >> >> >> To me that makes sense. Normally the memory installed also scales with the >> >> number of cores. >> >> >> >> Are you saying that we should look into modifying that scaling factor in >> >> perf_mmap()? Or that we should still add something to userspace for >> >> coresight to limit user supplied buffer sizes? >> > >> > I don't think we should modify the scaling factor in perf_mmap(), the >> > logic is not only used by AUX buffer, it's shared by normal event >> > ring buffer. >> > >> >> I think it makes sense to allow the user to specify any value that will work, >> >> it's up to them. >> > >> > Understand, I verified this patch with below steps: >> > >> > root@debian:~# echo 0 > /proc/sys/kernel/perf_event_paranoid >> > >> > leoy@debian:~$ perf record -e cs_etm// -m 4M,8M -o perf_test.data -- sleep 1 >> > Permission error mapping pages. >> > Consider increasing /proc/sys/kernel/perf_event_mlock_kb, >> > or try again with a smaller value of -m/--mmap_pages. >> > (current value: 1024,2048) >> > >> > leoy@debian:~$ perf record -e cs_etm// -m 4M,4M -o perf_test.data -- sleep 1 >> > Couldn't synthesize bpf events. >> > [ perf record: Woken up 1 times to write data ] >> > [ perf record: Captured and wrote 0.607 MB perf_test.data ] >> > >> > So this patch looks good for me: >> > >> > Reviewed-by: Leo Yan <leo.yan@linaro.org> >> > >> Thanks Leo! > >Arnaldo is not on the recipient list and as such he won't see this patch... > I saw it now, can I take this as an acked-by: Matthieu too? - Arnaldo
Em Thu, Dec 09, 2021 at 02:16:43PM +0000, James Clark escreveu: > > > On 09/12/2021 13:44, Leo Yan wrote: > > On Wed, Dec 08, 2021 at 02:08:04PM +0000, James Clark wrote: > >> On 08/12/2021 13:17, Leo Yan wrote: > >>> Hi James, > >>> > >>> On Wed, Dec 08, 2021 at 11:54:35AM +0000, James Clark wrote: > >>>> There are two checks, one is for size when running without admin, but > >>>> this one is covered by the driver and reported on in more detail here > >>>> (builtin-record.c): > >>>> > >>>> pr_err("Permission error mapping pages.\n" > >>>> "Consider increasing " > >>>> "/proc/sys/kernel/perf_event_mlock_kb,\n" > >>>> "or try again with a smaller value of -m/--mmap_pages.\n" > >>>> "(current value: %u,%u)\n", > >>> > >>> I looked into the kernel code and found: > >>> > >>> sysctl_perf_event_mlock = 512 + (PAGE_SIZE / 1024); // 512KB + 1 page > >>> > >>> If the system have multiple cores, let's say 8 cores, then kernel even > >>> can relax the limitaion with: > >>> > >>> user_lock_limit *= num_online_cpus(); > >>> > >>> So means the memory lock limitation is: > >>> > >>> (512KB + 1 page) * 8 = 4MB + 8 pages. > >>> > >>> Seems to me, it's much relax than the user space's limitaion 128KB. > >>> And let's imagine for Arm server, the permitted buffer size can be a > >>> huge value (e.g. for a system with 128 cores). > >>> > >>> Could you confirm if this is right? > >> > >> Yes that seems to be the case. And the commit message for that addition > >> states the reasoning: > >> > >> perf_counter: Increase mmap limit > >> > >> In a default 'perf top' run the tool will create a counter for > >> each online CPU. With enough CPUs this will eventually exhaust > >> the default limit. > >> > >> So scale it up with the number of online CPUs. > >> > >> To me that makes sense. Normally the memory installed also scales with the > >> number of cores. > >> > >> Are you saying that we should look into modifying that scaling factor in > >> perf_mmap()? Or that we should still add something to userspace for > >> coresight to limit user supplied buffer sizes? > > > > I don't think we should modify the scaling factor in perf_mmap(), the > > logic is not only used by AUX buffer, it's shared by normal event > > ring buffer. > > > >> I think it makes sense to allow the user to specify any value that will work, > >> it's up to them. > > > > Understand, I verified this patch with below steps: > > > > root@debian:~# echo 0 > /proc/sys/kernel/perf_event_paranoid > > > > leoy@debian:~$ perf record -e cs_etm// -m 4M,8M -o perf_test.data -- sleep 1 > > Permission error mapping pages. > > Consider increasing /proc/sys/kernel/perf_event_mlock_kb, > > or try again with a smaller value of -m/--mmap_pages. > > (current value: 1024,2048) > > > > leoy@debian:~$ perf record -e cs_etm// -m 4M,4M -o perf_test.data -- sleep 1 > > Couldn't synthesize bpf events. > > [ perf record: Woken up 1 times to write data ] > > [ perf record: Captured and wrote 0.607 MB perf_test.data ] > > > > So this patch looks good for me: > > > > Reviewed-by: Leo Yan <leo.yan@linaro.org> > > > Thanks Leo! Thanks, applied. - Arnaldo
diff --git a/tools/perf/arch/arm/util/cs-etm.c b/tools/perf/arch/arm/util/cs-etm.c index 293a23bf8be3..8a3d54a86c9c 100644 --- a/tools/perf/arch/arm/util/cs-etm.c +++ b/tools/perf/arch/arm/util/cs-etm.c @@ -407,25 +407,6 @@ static int cs_etm_recording_options(struct auxtrace_record *itr, } - /* Validate auxtrace_mmap_pages provided by user */ - if (opts->auxtrace_mmap_pages) { - unsigned int max_page = (KiB(128) / page_size); - size_t sz = opts->auxtrace_mmap_pages * (size_t)page_size; - - if (!privileged && - opts->auxtrace_mmap_pages > max_page) { - opts->auxtrace_mmap_pages = max_page; - pr_err("auxtrace too big, truncating to %d\n", - max_page); - } - - if (!is_power_of_2(sz)) { - pr_err("Invalid mmap size for %s: must be a power of 2\n", - CORESIGHT_ETM_PMU_NAME); - return -EINVAL; - } - } - if (opts->auxtrace_snapshot_mode) pr_debug2("%s snapshot size: %zu\n", CORESIGHT_ETM_PMU_NAME, opts->auxtrace_snapshot_size);
There are two checks, one is for size when running without admin, but this one is covered by the driver and reported on in more detail here (builtin-record.c): pr_err("Permission error mapping pages.\n" "Consider increasing " "/proc/sys/kernel/perf_event_mlock_kb,\n" "or try again with a smaller value of -m/--mmap_pages.\n" "(current value: %u,%u)\n", This had the effect of artificially limiting the aux buffer size to a value smaller than what was allowed because perf_event_mlock_kb wasn't taken into account. The second is to check for a power of two, but this is covered here (evlist.c): pr_info("rounding mmap pages size to %s (%lu pages)\n", buf, pages); Signed-off-by: James Clark <james.clark@arm.com> --- tools/perf/arch/arm/util/cs-etm.c | 19 ------------------- 1 file changed, 19 deletions(-)