On Mon, Nov 11, 2019 at 3:45 AM Andreas Schwab <[email protected]> wrote:
> Only ubsan is supported so far. This has been tested on openSUSE
> Tumbleweed, there are no testsuite failures.
>
> * configure.tgt (riscv64-*-linux*): Enable build.
I tried this on my riscv64 Fedora system, and I get a build error.
In file included from
../../../../gcc-git/libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:167:
../../../../gcc-git/libsanitizer/sanitizer_common/sanitizer_internal_defs.h:342:72:
error: narrowing conversion of ‘-1’ from ‘int’ to ‘long unsigned int’
[-Wnarrowing]
342 | typedef char IMPL_PASTE(assertion_failed_##_, line)[2*(int)(pred)-1]
| ^
...
../../../../gcc-git/libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.cpp:1136:1:
note: in expansion of macro ‘CHECK_SIZE_AND_OFFSET’
1136 | CHECK_SIZE_AND_OFFSET(ipc_perm, mode);
| ^~~~~~~~~~~~~~~~~~~~~
It appears that the problem is that my system defines the icp_perm
mode field as __mode_t whereas the sanitizer assumes it will be
unsigned short, followed by an unsigned short pad field. I haven't
looked at the full history yet, but there were apparently similar
problems with the aarch64 port, so maybe we need some special code for
riscv to make this work? I don't know why it worked for you. Maybe a
different glibc or kernel version?
Incidentally, the
libsanitizer/sanitizer_common/sanitizer_platform_limits_posix.h file
has an obvious problem in the struct __sanitizer_ipc_perm definition,
because it has an ifdef for __aarch64__ inside an ifdef for __sparc__,
and there is no way the __aarch64__ test can ever succeed there.
There is a second __aarch64__ test a little farther down that looks
OK. Maybe a patch merge error?
Jim
