Summary: | [GTK] Layout Tests imported/w3c/web-platform-tests/WebCryptoAPI/derive_bits_keys/ecdh_bits.worker.html crashes | ||||||
---|---|---|---|---|---|---|---|
Product: | WebKit | Reporter: | Fujii Hironori <Hironori.Fujii> | ||||
Component: | WebKitGTK | Assignee: | Nobody <webkit-unassigned> | ||||
Status: | RESOLVED DUPLICATE | ||||||
Severity: | Normal | CC: | bugs-noreply, zan | ||||
Priority: | P2 | ||||||
Version: | WebKit Nightly Build | ||||||
Hardware: | Unspecified | ||||||
OS: | Unspecified | ||||||
Attachments: |
|
Description
Fujii Hironori
2017-06-26 00:33:41 PDT
(In reply to Fujii Hironori from comment #0) > > STDERR: Ohhhh jeeee: MPI subsystem not initialized > > STDERR: fatal error in libgcrypt, file /home/slave/webkitgtk/gtk-linux-64-release-tests/build/WebKitBuild/DependenciesGTK/Source/libgcrypt-1.7.6/src/misc.c, line 140, function _gcry_logv: internal error (fatal or bug) This message is output in _gcry_mpi_const of WebKitBuild/DependenciesGTK/Source/libgcrypt-1.7.6/mpi/mpiutil.c. This means _gcry_mpi_init doesn's seem to be called. _gcry_mpi_init is called from global_init of libgcrypt-1.7.6/src/global.c. global_init has this comment. > /* This is our handmade constructor. It gets called by any function > likely to be called at startup. The suggested way for an > application to make sure that this has been called is by using > gcry_check_version. */ gcry_check_version() needs to be called to initialize at startup. Created attachment 313821 [details] WIP patch This WIP patch solves the crashes. But, two problems. * I'm not sure this is the right place to call gcry_check_version(). * Solved Crash problem. But, the test cases still Failed. I think Bug 169272 looks the same. Zan, do you have any idea? Yes, libgcrypt has to be explicitly initialized. Otherwise the test relies on being run after other tests that force through the necessary initialization. I'm handling this in bug #173589. The ec_importKey.worker.html and test_ec_importKey.https.html tests still fail due to missing implementations. TestExpectations address that, but not the crashes. Thank you. Closed this bug. *** This bug has been marked as a duplicate of bug 173589 *** |