Re: [PATCH v2 12/12] linux-user: fix clock_nanosleep()

[Laurent Vivier]

Le 22/07/2020 à 08:49, Laurent Vivier a écrit :
> Le 22/07/2020 à 08:29, Alex Bennée a écrit :
>> From: Laurent Vivier <laurent@vivier.eu>
>>
>> If clock_nanosleep() encounters an error, it returns one of the positive
>> error number.
>>
>> If the call is interrupted by a signal handler, it fails with error EINTR
>> and if "remain" is not NULL and "flags" is not TIMER_ABSTIME, it returns
>> the remaining unslept time in "remain".
>>
>> Update linux-user to not overwrite the "remain" structure if there is no
>> error.
>>
>> Found with "make check-tcg", linux-test fails on nanosleep test:
>>
>>   TEST    linux-test on x86_64
>> .../tests/tcg/multiarch/linux-test.c:242: nanosleep
>> make[2]: *** [../Makefile.target:153: run-linux-test] Error 1
>> make[1]: *** [.../tests/tcg/Makefile.qemu:76: run-guest-tests] Error 2
>> make: *** [.../tests/Makefile.include:857: run-tcg-tests-x86_64-linux-user] 
>> Error 2
>>
>> Reported-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>> Signed-off-by: Laurent Vivier <laurent@vivier.eu>
>> Message-Id: <20200721201754.2731479-1-laurent@vivier.eu>
>> ---
>>  linux-user/syscall.c | 15 ++++++++++++---
>>  1 file changed, 12 insertions(+), 3 deletions(-)
>>
>> diff --git a/linux-user/syscall.c b/linux-user/syscall.c
>> index 1211e759c26..caa7cd3cab9 100644
>> --- a/linux-user/syscall.c
>> +++ b/linux-user/syscall.c
>> @@ -11829,10 +11829,19 @@ static abi_long do_syscall1(void *cpu_env, int 
>> num, abi_long arg1,
>>      {
>>          struct timespec ts;
>>          target_to_host_timespec(&ts, arg3);
>> -        ret = get_errno(safe_clock_nanosleep(arg1, arg2,
>> -                                             &ts, arg4 ? &ts : NULL));
>> -        if (arg4)
>> +        /*
>> +         * clock_nanosleep() returns 0 or one of the *positive* error 
>> number.
>> +         */
>> +        ret = host_to_target_errno(safe_clock_nanosleep(arg1, arg2, &ts,
>> +                                                        arg4 ? &ts : NULL));
>> +        /*
>> +         * if the call is interrupted by a signal handler, it fails
>> +         * with error TARGET_EINTR and if arg4 is not NULL and arg2 is not
>> +         * TIMER_ABSTIME, it returns the remaining unslept time in arg4.
>> +         */
>> +        if (ret == TARGET_EINTR && arg4 && arg2 != TIMER_ABSTIME) {
>>              host_to_target_timespec(arg4, &ts);
>> +        }
>>  
>>  #if defined(TARGET_PPC)
>>          /* clock_nanosleep is odd in that it returns positive errno values.
>>
> 
> Wait a little before pushing that: I've made more tests and it seems to
> break something in LTP. I have to analyze.

Apparently our safe_clock_nanosleep() doesn't behave like the system one
described in the manpage: it actually returns -1 and update errno.
So we need to keep the get_errno() and I think TARGET_PPC part can be
removed because the crf bit will be updated in ppc/cpu_loop.c.

I update and test my patch and I will send the v2.

Thanks,
Laurent