[compiler-rt] r300150 - [msan] Fix msan_test broken after r299884.

[Vitaly Buka via llvm-commits]

reverted r300249, new fix r300250

On Thu, Apr 13, 2017 at 11:20 AM Vitaly Buka <vitalybuka at google.com> wrote:

> Thanks, I'll update the test.
>
> On Thu, Apr 13, 2017 at 6:55 AM Hal Finkel <hfinkel at anl.gov> wrote:
>
>> Hi Vitaly,
>>
>> With further testing, I recommend that you revert this. It does not
>> actually solve the problem. In short, I now see this failure mode:
>>
>>      MemorySanitizer-Unit ::
>> Msan-x86_64-Test/IpTests/MemorySanitizerIpTest.accept/1
>>      MemorySanitizer-Unit ::
>> Msan-x86_64-Test/IpTests/MemorySanitizerIpTest.recvmsg/1
>>      MemorySanitizer-Unit ::
>> Msan-x86_64-with-call-Test/IpTests/MemorySanitizerIpTest.accept/1
>>      MemorySanitizer-Unit ::
>> Msan-x86_64-with-call-Test/IpTests/MemorySanitizerIpTest.recvmsg/1
>>
>> [ RUN      ] IpTests/MemorySanitizerIpTest.accept/1
>>
>> /home/hfinkel/public/src/llvm/projects/compiler-rt/lib/msan/tests/msan_test.cc:990:
>> Failure
>>        Expected: 115
>> To be equal to: (*__errno_location ())
>>        Which is: 113
>> [  FAILED  ] IpTests/MemorySanitizerIpTest.accept/1, where GetParam() =
>> 10 (4 ms)
>> [----------] 1 test from IpTests/MemorySanitizerIpTest (4 ms total)
>>
>> Meaning that, instead of returning EINPROGRESS, it returns EHOSTUNREACH
>> (which is consistent with the lack of IPv6 on the interfaces).
>>
>> ...
>>
>> [ RUN      ] IpTests/MemorySanitizerIpTest.recvmsg/1
>>
>> /home/hfinkel/public/src/llvm/projects/compiler-rt/lib/msan/tests/msan_test.cc:1046:
>> Failure
>> Expected: (0) < (res), actual: 0 vs -1
>> [  FAILED  ] IpTests/MemorySanitizerIpTest.recvmsg/1, where GetParam() =
>> 10 (4 ms)
>> [----------] 1 test from IpTests/MemorySanitizerIpTest (4 ms total)
>>
>> ...
>>
>> (i.e. sendmsg did not actually send any data - we don't report errno in
>> this case).
>>
>> Instead of making the tests robust to all of the ways that the socket
>> calls might fail on a system that does not actually have IPv6 configured
>> (even if the kernel supports it), I suggest just asking for the loopback
>> interface and then dealing with the fact that the bind call might fail.
>> On my system, binding to the loopback when it has no IPv6 assignment
>> sets errno to EADDRNOTAVAIL.
>>
>> Thanks again,
>> Hal
>>
>> On 04/12/2017 07:36 PM, Vitaly Buka via llvm-commits wrote:
>> > Author: vitalybuka
>> > Date: Wed Apr 12 19:36:03 2017
>> > New Revision: 300150
>> >
>> > URL: http://llvm.org/viewvc/llvm-project?rev=300150&view=rev
>> > Log:
>> > [msan] Fix msan_test broken after r299884.
>> >
>> > Bind to ANY as some machines may have IPv6 support but without IPv6 on
>> loopback
>> > interface.
>> >
>> > Reviewers: eugenis
>> >
>> > Subscribers: llvm-commits
>> >
>> > Differential Revision: https://reviews.llvm.org/D31998
>> >
>> > Modified:
>> >      compiler-rt/trunk/lib/msan/tests/msan_test.cc
>> >
>> > Modified: compiler-rt/trunk/lib/msan/tests/msan_test.cc
>> > URL:
>> http://llvm.org/viewvc/llvm-project/compiler-rt/trunk/lib/msan/tests/msan_test.cc?rev=300150&r1=300149&r2=300150&view=diff
>> >
>> ==============================================================================
>> > --- compiler-rt/trunk/lib/msan/tests/msan_test.cc (original)
>> > +++ compiler-rt/trunk/lib/msan/tests/msan_test.cc Wed Apr 12 19:36:03
>> 2017
>> > @@ -899,7 +899,7 @@ class SocketAddr4 : public SocketAddr {
>> >       memset(&sai_, 0, sizeof(sai_));
>> >       sai_.sin_family = AF_INET;
>> >       sai_.sin_port = port;
>> > -    sai_.sin_addr.s_addr = htonl(INADDR_LOOPBACK);
>> > +    sai_.sin_addr.s_addr = htonl(INADDR_ANY);
>> >     }
>> >
>> >     sockaddr *ptr() override { return reinterpret_cast<sockaddr
>> *>(&sai_); }
>> > @@ -917,7 +917,7 @@ class SocketAddr6 : public SocketAddr {
>> >       memset(&sai_, 0, sizeof(sai_));
>> >       sai_.sin6_family = AF_INET6;
>> >       sai_.sin6_port = port;
>> > -    sai_.sin6_addr = in6addr_loopback;
>> > +    sai_.sin6_addr = in6addr_any;
>> >     }
>> >
>> >     sockaddr *ptr() override { return reinterpret_cast<sockaddr
>> *>(&sai_); }
>> >
>> >
>> > _______________________________________________
>> > llvm-commits mailing list
>> > llvm-commits at lists.llvm.org
>> > http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-commits
>>
>> --
>> Hal Finkel
>> Lead, Compiler Technology and Programming Languages
>> Leadership Computing Facility
>> Argonne National Laboratory
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20170413/8b18910b/attachment.html>