rathann

Great catch, @animalize . The following patch makes all tests pass on s390x: ``` diff diff -up zstandard-0.15.1/c-ext/compressionreader.c.orig zstandard-0.15.1/c-ext/compressionreader.c --- zstandard-0.15.1/c-ext/compressionreader.c.orig 2020-12-31 23:18:06.000000000 +0000 +++ zstandard-0.15.1/c-ext/compressionreader.c 2021-03-01 11:41:09.056060857 +0000 @@...

It gets assigned values 0 and 1, so I guess it doesn't matter much, as long as the type is used consistently everywhere. I think `char` is signed on s390x...

FWIW, the Fedora package has been carrying the merged patch since October last year and s390x builds had their test suites pass.

New ones in 1.0.33 (tested on Fedora 38+). I'm not sure why they occur, because I don't see any difference in the expected and actual outputs. ``` - test/wasm2c/spec/simd_splat.txt expected...

Same in 1.0.32. Note that I'll stop testing on ARM 32-bit in about 6 months, when Fedora 36 becomes EOL.

This is CVE-2023-27115 .

Two new test failures in 1.0.33 test suite on i686: ``` - test/wasm2c/spec/memory64/float_memory64.txt expected error code 0, got 1. STDERR MISMATCH: --- expected +++ actual @@ -0,0 +1,22 @@ +Traceback...

> I think the root cause might be something like, "The SSE2 floating-point instructions largely conform to IEEE 754 (and the Wasm semantics wrt canonical/arithmetic NaNs), but x87 doesn't. [...]...

> It's cool that Fedora's building WABT with those CFLAGS, but to get the wasm2c tests to build _their_ output with those CFLAGS, I think they have to be added...

How about simply "comment" instead of "vote"? Comments can be positive and negative, too.