[LTP] [PATCH v3] syscalls/sync_file_range: add partial file sync test-cases

[Sumit Garg]

On Mon, 10 Jun 2019 at 09:03, Li Wang <liwang@redhat.com> wrote:
>
>
>
> On Thu, Mar 28, 2019 at 12:57 PM Sumit Garg <sumit.garg@linaro.org> wrote:
>>
>> On Wed, 27 Mar 2019 at 20:18, Cyril Hrubis <chrubis@suse.cz> wrote:
>> >
>> > Hi!
>> > Sorry for the long delay.
>> >
>> > This is altmost perfect, the only problem is that the third test fails
>> > on vfat. As far as I can tell the reason is that vfat does not support
>> > sparse files, hence seeking to the middle of file and writing there also
>> > schedulles I/O to write zeros from the start of the file to the offset
>> > we started writing to.
>> >
>>
>> Hmm, I see.
>>
>> > Following ugly patch solves the problem:
>> >
>> > diff --git a/testcases/kernel/syscalls/sync_file_range/sync_file_range02.c b/testcases/kernel/syscalls/sync_file_range/sync_file_range02.c
>> > index 334ea5e88..774524c2f 100644
>> > --- a/testcases/kernel/syscalls/sync_file_range/sync_file_range02.c
>> > +++ b/testcases/kernel/syscalls/sync_file_range/sync_file_range02.c
>> > @@ -45,6 +45,12 @@ static void verify_sync_file_range(struct testcase *tc)
>> >
>> >         fd = SAFE_OPEN(tc->fname, O_RDWR|O_CREAT, MODE);
>> >
>> > +       if (!strcmp(tst_device->fs_type, "vfat")) {
>> > +               tst_res(TINFO, "Pre-filling file");
>> > +               tst_fill_fd(fd, 0, tc->write_off, 1);
>> > +               fsync(fd);
>> > +       }
>> > +
>> >         lseek(fd, tc->write_off, SEEK_SET);
>> >
>> >
>> > So either we limit the tests so that the sync region does not overlap with the
>> > possible hole at the start of the file and loose some test coverage.
>> >
>> > Or we can add a function to the test library that would return true/false if
>> > sparse files are supported for a given FS.
>> >
>>
>> My initial thought behind this test-case was to run sync over a range
>> which is partially written. The other partial region not being written
>> could either be a hole or already synced data. So pre-fill file in
>> case of vfat looks sane option, but how about if we add pre-fill as
>> part of setup? Something like:
>
>
> I think this is a bit better. Could u send a new patch version?
>

Sure. BTW, apologies for the delay as I was busy with some other tasks.

-Sumit

>>
>> --- a/testcases/kernel/syscalls/sync_file_range/sync_file_range02.c
>> +++ b/testcases/kernel/syscalls/sync_file_range/sync_file_range02.c
>> @@ -86,6 +86,12 @@ static void setup(void)
>>  {
>>         if (!check_sync_file_range())
>>                 tst_brk(TCONF, "sync_file_range() not supported");
>> +
>> +       if (!strcmp(tst_device->fs_type, "vfat")) {
>> +               tst_res(TINFO, "Pre-filling file");
>> +               tst_fill_file(FNAME3, 0, TST_MB, FILE_SZ_MB);
>> +               sync();
>> +       }
>>  }
>>
>> -Sumit
>>
>> > --
>> > Cyril Hrubis
>> > chrubis@suse.cz
>>
>> --
>> Mailing list info: https://lists.linux.it/listinfo/ltp
>
>
>
> --
> Regards,
> Li Wang