https://sourceware.org/bugzilla/show_bug.cgi?id=35009
--- Comment #2 from Aurelien Jarno <aurelien at aurel32 dot net> --- (In reply to Jessica Clarke from comment #1) > Hm, I'm confused by the binaries. > libgallium-26.2.6-1.so.gdb-18-branchpoint-629-g2c4ccb8103d looks like it was > built with -mtls-dialect=desc but a binutils that doesn't do relaxation, and > libgallium-26.2.6-1.so.gdb-18-branchpoint-628-g60b08e6b3ad looks like it was > built with -mtls-dialect=trad (and which binutils doesn't matter), given the > former has R_RISCV_TLSDESC relocations and the latter R_RISCV_DTPMOD64 > relocations in 1:1 correspondence (plus an extra R_RISCV_JUMP_SLOT for the > latter to call __tls_get_addr). Hmm indeed, your right, I didn't notice when doing the bisect that the configure output changed: - With binutils 2.27, the configure script says: Checking if "-mtls-dialect=gnu2" runs: DID NOT COMPILE Checking if "-mtls-dialect=desc" runs: NO (127) (see https://buildd.debian.org/status/fetch.php?pkg=mesa&arch=riscv64&ver=26.2.5-1&stamp=1792602107&raw=0 ) - With binutils with TLSDESC support, it says: Checking if "-mtls-dialect=gnu2" runs: DID NOT COMPILE Checking if "-mtls-dialect=desc" runs: YES (see https://buildd.debian.org/status/fetch.php?pkg=mesa&arch=riscv64&ver=26.2.6-1&stamp=1793899311&raw=0) Now the question is to know from where the problems come from. -- You are receiving this mail because: You are on the CC list for the bug.
