quic-garvgupt wrote: > > "Gentle ping @petrhosek @smithp35 - if the changes look good and all your > > comments have been addressed, can you approve the PR as well. Thanks!" > > Apologies for the delay. I've been away for the past couple of weeks on > holiday and am just catching up. > > I don't have any objections to the patch. I'm happy for this to go through > using the existing approval. I'm confident that this won't adversely affect > the LLVM bare-metal driver using compiler-rt, but I can't offer much > experience with libgcc. > > The one area I was concerned about was the lack of crtbeginS.o on the > `aarch64-none-elf` and `arm-none-eabi` GNU toolchains. Looking into this, > they have been configured with `--disable-shared` > https://gitlab.arm.com/tooling/gnu-devtools-for-arm/-/blob/main/build-baremetal-toolchain.sh?ref_type=heads#L1157 > so they just use the crtbegin.o and the GNU ld default -z notext even when > static-pie is used. When I try using --static-pie on these toolchains the > resulting binary is static but it is not PIE. > > I think using crtbeginS in LLVM is fine as it is probably best that there is > a link error for static-pie to catch the non-PIE nature of the output.
Thanks Peter for providing the information from the GNU toolchain side. I also think that if `static-pie` is used then it is not prudent to generate static binary, atleast not without any warning to user. Thanks for the approval. I will merge the changes. https://github.com/llvm/llvm-project/pull/147589 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
