Re: setitimer() doesn't work properly starting with 4.8-rc1

[Thomas Gleixner <tglx@linutronix.de>]

On Wed, 17 Aug 2016, Andrei Vagin wrote:

Cc'ing scheduler folks.

> On Wed, Aug 17, 2016 at 12:01 PM, Andrei Vagin <avagin@gmail.com> wrote:
> > Hello,
> >
> > I found that setitimer() doesn't work properly on out test AMD fitlet.
> > It works correctly in the 4.7 kernel and doesn't work in 4.8-rc1.
> >
> > A small test is attached to this message.
> >
> > [root@usr-LAN-dhcp-38 ~]# uname -a
> > Linux usr-LAN-dhcp-38.99.sw.ru 4.8.0-0.rc2.git1.1.fc25.x86_64 #1 SMP
> > Tue Aug 16 22:08:16 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux
> > [root@usr-LAN-dhcp-38 ~]# strace ./test-timers
> > ...
> > rt_sigaction(SIGVTALRM, {0x400606, [VTALRM], SA_RESTORER|SA_RESTART,
> > 0x7f51e0d67770}, {SIG_DFL, [], 0}, 8) = 0
> > setitimer(ITIMER_VIRTUAL, {it_interval={0, 100000}, it_value={0,
> > 100}}, NULL) = 0
> > exit_group(0)                           = ?
> > +++ exited with 0 +++
> 
> One more interesting thing is how cpu time is accounted. The test is
> running in a busy loop for 10 seconds, but the user time is zero and
> everything is accounted to the system time.
> 
> [root@usr-LAN-dhcp-38 ~]# time ./test-timers
> 
> real 0m10.959s
> user 0m0.000s
> sys 0m10.682s
> 
> 
> >
> > We can see that a test didn't get any signals. When timers works, this
> > output looks like this:
> > rt_sigaction(SIGALRM, {0x400606, [ALRM], SA_RESTORER|SA_RESTART,
> > 0x7f9e21b1b770}, {SIG_DFL, [], 0}, 8) = 0
> > setitimer(ITIMER_REAL, {it_interval={0, 100000}, it_value={0, 100}}, NULL) = 0
> > --- SIGALRM {si_signo=SIGALRM, si_code=SI_KERNEL} ---
> > rt_sigreturn({mask=[]})                 = 6295208
> > --- SIGALRM {si_signo=SIGALRM, si_code=SI_KERNEL} ---
> > rt_sigreturn({mask=[]})                 = 0
> > --- SIGALRM {si_signo=SIGALRM, si_code=SI_KERNEL} ---
> > rt_sigreturn({mask=[]})                 = 1
> > --- SIGALRM {si_signo=SIGALRM, si_code=SI_KERNEL} ---
> > rt_sigreturn({mask=[]})                 = 1
> > --- SIGALRM {si_signo=SIGALRM, si_code=SI_KERNEL} ---
> > rt_sigreturn({mask=[]})                 = 1471459909
> > --- SIGALRM {si_signo=SIGALRM, si_code=SI_KERNEL} ---
> > rt_sigreturn({mask=[]})                 = 1
> > --- SIGALRM {si_signo=SIGALRM, si_code=SI_KERNEL} ---
> >
> >
> > Only ITIMER_PROF and ITIMER_VIRTUAL doesn't work properly. ITIMER_REAL
> > works as expected.
> >
> > [root@usr-LAN-dhcp-38 ~]# cat /proc/cpuinfo
> > processor : 0
> > vendor_id : AuthenticAMD
> > cpu family : 22
> > model : 48
> > model name : AMD A10 Micro-6700T APU+AMD Radeon R6 Graphics
> > stepping : 1
> > microcode : 0x7030105
> > cpu MHz : 1000.000
> > cache size : 2048 KB
> > physical id : 0
> > siblings : 4
> > core id : 0
> > cpu cores : 4
> > apicid : 0
> > initial apicid : 0
> > fpu : yes
> > fpu_exception : yes
> > cpuid level : 13
> > wp : yes
> > flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov
> > pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt
> > pdpe1gb rdtscp lm constant_tsc rep_good acc_power nopl nonstop_tsc
> > extd_apicid aperfmperf eagerfpu pni pclmulqdq monitor ssse3 cx16
> > sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm
> > cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch
> > osvw ibs skinit wdt topoext perfctr_nb bpext ptsc perfctr_l2 cpb
> > hw_pstate vmmcall bmi1 xsaveopt arat npt lbrv svm_lock nrip_save
> > tsc_scale flushbyasid decodeassists pausefilter pfthreshold
> > overflow_recov
> > bugs : fxsave_leak sysret_ss_attrs null_seg
> > bogomips : 2395.40
> > TLB size : 1024 4K pages
> > clflush size : 64
> > cache_alignment : 64
> > address sizes : 40 bits physical, 48 bits virtual
> > power management: ts ttp tm 100mhzsteps hwpstate cpb acc_power [13]
> >
> > This problem was caught by criu tests:
> > https://github.com/xemul/criu/issues/200
> >
> > Thanks,
> > Andrei
>