RFR(S): 8252407: Build failure with gcc-8+ and asan

Eric Liu eric.c.liu at arm.com
Wed Sep 2 13:40:17 UTC 2020


Hi David,

Thanks for your review and sorry for the delay.

David Holmes <david.holmes at oracle.com> on Tue Sep 1 08:57:01 UTC 2020

> We already build with gcc 9.2 and don't see any issues with shared code 
> compiling for hotspot - which of these require --enable-asan? It 
> certainly looks like an unnecessary warning - the whole point of strncpy 
> is to truncate at a safe length. :(

I suppose all of those warnings require "--enable-asan", but two of them (test related)
also require "--with-gtest".

> Have you tested this by building on other platforms? I'm not sure if all 
> compilers will ignore these gcc specific pragmas.

Yes. Besides aarch64, I also tested this by building on x86. But a harder question
now is the different behaviors between gcc-9 and gcc-10. As Kim Barrett<kim.barrett at oracle.com>
said it would result to a warning cycle here.

Thanks,
Eric


More information about the serviceability-dev mailing list