mips test failures

[Dmitry V. Levin]

On Mon, Mar 02, 2015 at 07:26:08PM -0500, Mike Frysinger wrote:
> i got a mips box online that can do one-off o32/n32/n64 testing.  i don't have 
> it fully hooked up yet into the rest of my build scripts yet.  it's n32 native
> though, so that might cause some issues ?  not sure how many of these are
> regressions.  at least each builds cleanly ? :)

Regressions?  This is the first time I hear about native mips n32 build.
That's a clear progress! :)

> testing d50949d21c2b07daadc26c7302163dc6066d4a75
> 
> mips o32:
> FAIL: getdents.test
> FAIL: getrandom.test
> FAIL: stat.test
> FAIL: stat32-v.test
> FAIL: uid.test
> FAIL: count.test
> 
> mips n64:
> FAIL: getdents.test
> FAIL: stat.test
> FAIL: count.test

With all syscalls prefixed with "n32_" I'd expect most of tests fail.
I'm very much surprised that so many tests actually passed.

> mips n32:
> FAIL: stat32-v.test
> stat\("stat32_sample", \{st_dev=makedev\(8, 4\), st_ino=3296092, st_mode=0\|0, st_nlink=33188, st_uid=1, st_gid=8282, st_blksize=1425341901, st_blocks=547422357, st_size=0, st_atime=1977/05/29-23:33:01, st_mtime=2001/02/03-04:05:00, st_ctime=2001/02/03-04:05:00(, st_flags=[0-9]+)?(, st_fstype=[^,]*)?(, st_gen=[0-9]+)?\}\) += 0
> stat("stat32_sample", {st_dev=makedev(8, 4), st_ino=3296092, st_mode=S_IFREG|0644, st_nlink=1, st_uid=8282, st_gid=100, st_blksize=4096, st_blocks=0, st_size=233811181, st_atime=2001/02/03-04:05:00, st_mtime=2001/02/03-04:05:00, st_ctime=2015/03/02-19:18:21}) = 0

strace itself is correct while the test is not.

> FAIL: uio.test
> pread64(3, "\0\0\0\0", 4, 1004211379570065135) = 4
> preadv(3, [{"\0\0\0\0", 4}], 1, 18446744073150512879) = 4
> pwrite64(3, "\0\0\0\0", 4, 1004211379570065135) = 4
> pwritev(3, [{"\0\0\0\0", 4}], 1, 18446744073150512879) = 4

Decoding of preadv/pwritev offsets is wrong.
Must be a bug in print_llu_from_low_high_val().

> FAIL: strace-k.test

I have not the slightest idea whether unw_init_remote works on mips
or not.  Anyway, -k is experimental, this test hardly passes on any
architecture besides x86_64, so I wouldn't bother.


-- 
ldv
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: not available
URL: <http://lists.strace.io/pipermail/strace-devel/attachments/20150303/12d68fe5/attachment.bin>