On 03/10/13 14:02, Peter Zijlstra wrote:
> On Thu, Oct 03, 2013 at 01:13:45PM +0300, Adrian Hunter wrote:
>>> Anyway; looking at this, why does time_zero have these different checks
>>> from the other time bits?
>>>
>>> @@ -1897,6 +1898,11 @@ void arch_perf_update_userpage(struct
>>> perf_event_mmap_page *userpg, u64 now)
>>> userpg->time_mult = this_cpu_read(cyc2ns);
>>> userpg->time_shift = CYC2NS_SCALE_FACTOR;
>>> userpg->time_offset = this_cpu_read(cyc2ns_offset) - now;
>>> +
>>> + if (sched_clock_stable && !check_tsc_disabled()) {
>>> + userpg->cap_usr_time_zero = 1;
>>> + userpg->time_zero = this_cpu_read(cyc2ns_offset);
>>> + }
>>> }
>>>
>>> That doesn't make any kind of sense.. why is cyc2ns_offset differently
>>> tested from cyc2ns itself?
>>
>> I am afraid I don't understand the scaling calculations
>> so I don't know if they make any sense.
>>
>> cap_usr_time_zero (now cap_user_time_zero) means you can convert
>> perf time to / from TSC. That only works if TSC is not disabled
>> and sched_clock is stable (and you have constant, non-stop TSC)
>>
>> As far as I can tell, assuming the hardware is not broken,
>> sched_clock will be stable unless something (BIOS) or someone
>> (meddling user) has changed TSC manually.
>
> Well all the scaling mess only applies to TSC.. furthermore note how
> time_offset even uses cyc2ns_offset.
>
> So I don't see any point in cap_user_time and cap_user_time_zero having
> different conditions.
>
> Also, I don't think that a cpu without cap_tsc will ever run perf code
> so I don't see the point of testing tsc_disabled; if someone is mad
> enough to boot with that he can bloody well keep the pieces.
You can boot with notsc and perf still seems to work
e.g.
$ dmesg | grep -i tsc
[ 0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-3.11.0+
root=UUID=f1b4c71a-15aa-41a6-8898-cdde49966bce ro ignore_loglevel
earlyprintk=ttyS0 console=ttyS0,115200n8 no_console_suspend notsc=1
[ 0.000000] Kernel command line: BOOT_IMAGE=/boot/vmlinuz-3.11.0+
root=UUID=f1b4c71a-15aa-41a6-8898-cdde49966bce ro ignore_loglevel
earlyprintk=ttyS0 console=ttyS0,115200n8 no_console_suspend notsc=1
[ 0.000000] tsc: Kernel compiled with CONFIG_X86_TSC, cannot disable TSC
completely
[ 0.000000] tsc: Fast TSC calibration using PIT
[ 0.004000] tsc: Detected 2593.932 MHz processor
[ 0.224000] TSC deadline timer enabled
$ tools/perf/perf record -e instructions:u uname
Linux
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.032 MB perf.data (~1414 samples) ]
$ tools/perf/perf script
# ========
# captured on: Thu Oct 3 17:31:38 2013
# hostname : ivb01
# os release : 3.11.0+
# perf version : 3.11.rc4.gde8db8
# arch : x86_64
# nrcpus online : 8
# nrcpus avail : 8
# cpudesc : Intel(R) Core(TM) i7-3720QM CPU @ 2.60GHz
# cpuid : GenuineIntel,6,58,9
# total memory : 8074624 kB
# cmdline : /home/ahunter/git/linux-perf/tools/perf/perf record -e
instructions:u uname
# event : name = instructions:u, type = 0, config = 0x1, config1 = 0x0, config2
= 0x0, excl_usr = 0, excl_kern = 1, excl_host = 0, e
# HEADER_CPU_TOPOLOGY info available, use -I to display
# HEADER_NUMA_TOPOLOGY info available, use -I to display
# pmu mappings: cpu = 4, software = 1, tracepoint = 2, uncore_cbox_0 = 6,
uncore_cbox_1 = 7, uncore_cbox_2 = 8, uncore_cbox_3 = 9, b
# ========
#
uname 3914 451.484000: instructions:u: 7fe3f3315af3
[unknown] (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: ffffffff813a1ad0
page_fault ([kernel.kallsyms])
uname 3914 451.484000: instructions:u: 7fe3f3316136
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: ffffffff813a1ad0
page_fault ([kernel.kallsyms])
uname 3914 451.484000: instructions:u: ffffffff813a1ad0
page_fault ([kernel.kallsyms])
uname 3914 451.484000: instructions:u: ffffffff813a1ad0
page_fault ([kernel.kallsyms])
uname 3914 451.484000: instructions:u: 7fe3f33161b4
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: 7fe3f33161af
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: 7fe3f33161af
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: 7fe3f33161af
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: 7fe3f33161bd
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: 7fe3f3316388
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: 7fe3f3316590
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
uname 3914 451.484000: instructions:u: 7fe3f33165a0
_dl_start (/lib/x86_64-linux-gnu/ld-2.13.so)
>
> Arguably the CONSTANT && NONSTOP test was to test the same thing as
> sched_clock_stable, but I suppose we can use that.
>
> Which reminds me; I should go and fix the transition for
> sched_clock_stable: 1->0.
>
> ---
> diff --git a/arch/x86/kernel/cpu/perf_event.c
> b/arch/x86/kernel/cpu/perf_event.c
> index 897783b3302a..9db2b361a63d 100644
> --- a/arch/x86/kernel/cpu/perf_event.c
> +++ b/arch/x86/kernel/cpu/perf_event.c
> @@ -1888,21 +1888,16 @@ void arch_perf_update_userpage(struct
> perf_event_mmap_page *userpg, u64 now)
> userpg->cap_user_rdpmc = x86_pmu.attr_rdpmc;
> userpg->pmc_width = x86_pmu.cntval_bits;
>
> - if (!boot_cpu_has(X86_FEATURE_CONSTANT_TSC))
> - return;
> -
> - if (!boot_cpu_has(X86_FEATURE_NONSTOP_TSC))
> - return;
> + if (!sched_clock_stable)
> + return
>
> userpg->cap_user_time = 1;
> userpg->time_mult = this_cpu_read(cyc2ns);
> userpg->time_shift = CYC2NS_SCALE_FACTOR;
> userpg->time_offset = this_cpu_read(cyc2ns_offset) - now;
>
> - if (sched_clock_stable && !check_tsc_disabled()) {
> - userpg->cap_user_time_zero = 1;
> - userpg->time_zero = this_cpu_read(cyc2ns_offset);
> - }
> + userpg->cap_user_time_zero = 1;
> + userpg->time_zero = this_cpu_read(cyc2ns_offset);
> }
>
> /*
>
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
