This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: Does anyone want to delay creating the 2.31 release this weekend (July 7/8) ?
Hi H.J.,
> I don't know if static PIE should be fixed for MIPS:
>
> https://sourceware.org/bugzilla/show_bug.cgi?id=22263
> https://sourceware.org/bugzilla/show_bug.cgi?id=22570
Looking into it, thanks for the heads-up, but there is more to it as with
the fix applied (and updated to avoid a segfault with R_MIPS_TLS_LDM) I
get more dynamic relocations with a dynamic PDE than with a dynamic PIE,
both assembled from the same source, which is certainly not what one would
expect.
The relocations in question are R_MIPS_TLS_DTPMOD, R_MIPS_TLS_DTPREL and
R_MIPS_TLS_TPREL32 against a variable defined within the executable
itself, which should have been resolved at static link time. This comes
from a condition that looks like it has been inadvertently reversed in the
original commit 0f20cc3522e7 ("TLS support for MIPS"),
<https://sourceware.org/ml/binutils/2005-02/msg00607.html>, code, and then
(understandably, given the original breakage) incorrectly converted with
your commit 0e1862bb401f ("Add output_type to bfd_link_info").
I now have another fix to address that other problem, however given the
limited number of TLS tests we have I will rebuild and regression-test
glibc as well to be fully confident about that fix before going ahead with
it. Which means it may take a couple of days yet.
Maciej