bug#72239: libfaketime (still) broken on i686-linux

[Christopher Baines]

Ludovic Courtès <ludo@gnu.org> writes:

> Ludovic Courtès <ludovic.courtes@inria.fr> skribis:
>
>> So the easiest short-term solution seems to be using datefudge to run
>> the ‘nss’ tests on 32-bit platforms, as Chris already suggested before
>> (patch below; it’s being built right now, I’ll see tomorrow if it
>> worked…).
>
> It failed:
>
> --8<---------------cut here---------------start------------->8---
> $ gunzip < 
> /var/log/guix/drvs/yk/8dxp7v6v29nbhkyiv5x8sk04gszycm-nss-3.99.drv.gz | tail 
> -20
> NSS_DISABLE_HW_SHA2=
> NSS_DISABLE_PCLMUL=
> NSS_DISABLE_AVX=
> NSS_DISABLE_ARM_NEON=
> NSS_DISABLE_SSSE3=
>
> Tests summary:
> --------------
> Passed:             79015
> Failed:             2
> Failed with core:   0
> ASan failures:      0
> Unknown status:     2
> TinderboxPrint:Unknown: 2
>
> error: in phase 'check': uncaught exception:
> %exception #<&invoke-error program: "datefudge" arguments: ("2024-01-23" 
> "./nss/tests/all.sh") exit-status: 1 term-signal: #f stop-signal: #f>
> phase `check' failed after 12784.8 seconds
> command "datefudge" "2024-01-23" "./nss/tests/all.sh" failed with status 1
> build process 18 exited with status 256
>
> $ zgrep -B5 ' - FAILED$' 
> /var/log/guix/drvs/yk/8dxp7v6v29nbhkyiv5x8sk04gszycm-nss-3.99.drv.gz
> [  FAILED  ] Pkcs11RsaPssTestWycheproof.RsaPss4096Sha256
> [  FAILED  ] Pkcs11RsaPssTestWycheproof.RsaPss4096Sha512
> [  FAILED  ] Pkcs11RsaPssTestWycheproof.RsaPssMisc
>
> 39 FAILED TESTS
> gtests.sh: #54: pk11_gtest run successfully  - FAILED
> -
> [  FAILED  ] Pkcs11RsaPssTestWycheproof.RsaPss4096Sha256
> [  FAILED  ] Pkcs11RsaPssTestWycheproof.RsaPss4096Sha512
> [  FAILED  ] Pkcs11RsaPssTestWycheproof.RsaPssMisc
>
> 39 FAILED TESTS
> gtests.sh: #54: pk11_gtest run successfully  - FAILED
> --8<---------------cut here---------------end--------------->8---
>
> Not sure if it’s because datefudge doesn’t interpose on every symbol
> that we need or if there’s another problem.
>
> Thoughts?

So you can run just this failing gtests by adding (setenv "NSS_TESTS"
"gtests") to the check phase.

Then if you build with datefudge and without datefudge, there's some
difference in the output when the testsuite starts.

When running without datefudge, the tests work and the cd in gtests.sh
seems to fail:

Running tests for gtests
TIMESTAMP gtests BEGIN: Mon Jul 22 10:09:27 UTC 2024
./gtests.sh: line 25: cd: ./nss/tests: No such file or directory
/tmp/guix-build-nss-3.99.drv-0/nss-3.99/nss/tests/gtests

But when running with datefudge, the cd seems to work, but the tests
fail:

Running tests for gtests
TIMESTAMP gtests BEGIN: Tue Jan 23 00:00:01 UTC 2024
/tmp/guix-build-nss-3.99.drv-0/nss-3.99/nss/tests


This seems to relate to the test failures, as they seem to fail reading
a JSON file because they're using the wrong filename:

[----------] 1 test from Pkcs11AESKeyWrapKwpTest
[ RUN      ] Pkcs11AESKeyWrapKwpTest.TestVectors
json_reader.cc:13: Failure
Value of: f_
  Actual: false
Expected: true
error opening vectors from: 
/tmp/guix-build-nss-3.99.drv-0/nss-3.99/nss/tests/../../gtests/pk11_gtest/../common/testvectors/kwp-vectors.json
json_reader.cc:130: Failure
Failed
Unexpected '\0'
json_reader.cc:49: Failure
Expected equality of these values:
  take()
    Which is: '\0'
  ':'
    Which is: ':' (58, 0x3A)
pk11_aeskeywrapkwp_unittest.cc:120: Failure
Expected equality of these values:
  "numberOfTests"
  r.ReadLabel()
    Which is: ""
[  FAILED  ] Pkcs11AESKeyWrapKwpTest.TestVectors (0 ms)


There's two many .. here, it should be
/tmp/guix-build-nss-3.99.drv-0/nss-3.99/nss/tests/../gtests/pk11_gtest/../common/testvectors/kwp-vectors.json


Removing the cd line from the gtests.sh script seems to get the package
to build for i686-linux with datefudge, so I've sent a patch (#72244) to
this effect. It's odd that the cd command works but the tests fail
though.

signature.asc
Description: PGP signature