This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: [BFD][AARCH64]Disallow R_AARCH64_ABS32(LP64) & R_AARCH64_ABS16 in const section of shared object
- From: Matthias Klose <doko at ubuntu dot com>
- To: Renlin Li <renlin dot li at foss dot arm dot com>, "binutils at sourceware dot org" <binutils at sourceware dot org>, Nicholas Clifton <nickc at redhat dot com>, Ramana Radhakrishnan <Ramana dot Radhakrishnan at arm dot com>, Richard Earnshaw <Richard dot Earnshaw at arm dot com>
- Date: Fri, 2 Feb 2018 17:45:02 +0100
- Subject: Re: [BFD][AARCH64]Disallow R_AARCH64_ABS32(LP64) & R_AARCH64_ABS16 in const section of shared object
- Authentication-results: sourceware.org; auth=none
- References: <59E72FAB.6000804@foss.arm.com>
This seems to break kernel builds, and linking systemd against gnu-efi
libraries. Now reported in PR22764.
Matthias
On 18.10.2017 12:40, Renlin Li wrote:
> Hi all,
>
> R_AARCH64_ABS64, R_AARCH64_ABS32 and R_AARCH64_ABS16 are data relocations
> supported in AArch64 elf ABI.
>
> R_AARCH64_ABS64 under LP64 is allowed in shared object and a dynamic relocation
> entry
> will be generated. This allows the dynamic linker to do further symbol resolution.
>
> R_AARCH64_ABS32 likewise is allowed in shared object, however under ILP32 abi.
>
> The original behavior for R_AARCH64_ABS32 under LP64 is that, it's allowed
> in shared object and silently resolved at static linking time.
> No dynamic relocation entry is generate for it.
>
> R_AARCH64_ABS16 is allowed in shared object under both L64 and ILP32.
> It's resolved at static linking time as well.
>
> Under LP64, the address should be 64-bit. R_AARCH64_ABS32 relocation indicates
> an address that is only sized 32 bits which is not correct in LP64 shared object.
> It's useful to error out.
>
> I have checked glibc dynamic linker code, R_AARCH64_ABS16 is not supported at
> all. So
> R_AARCH64_ABS16 should be reject in shared object completely.
>
>
> In this patch, R_AARCH64_ABS32 is rejected under LP64 in read-only section of
> shared object.
> R_AARCH64_ABS16 is rejected in constant section of shared object in both ABI.
>
> This will sometimes provide useful information for buggy code.
>
> By the way, glibc aarch64 elf_machine_load_address is using R_AARCH64_ABS32 to
> hold an address even in LP64. I had a discussion here sometime ago. The change
> here will break the glibc build. This patch will not be committed until glibc
> has been fixed.
> https://sourceware.org/ml/libc-alpha/2017-10/msg00816.html
>
> testcases are also added. Binutils regression checked OK.
> Linux kernel linked OK with new binutils. OK for trunk?
>
>
> Regards,
> Renlin Li
>
>
> ld/ChangeLog:
>
> 2017-10-18 Renlin Li <renlin.li@arm.com>
>
> * testsuite/ld-aarch64/aarch64-elf.exp: Run new test cases.
> * testsuite/ld-aarch64/emit-relocs-258-dyn-bad.d: New.
> * testsuite/ld-aarch64/emit-relocs-258.s: New.
> * testsuite/ld-aarch64/emit-relocs-259-dyn-bad.d: New.
> * testsuite/ld-aarch64/emit-relocs-259.s: New.
> * testsuite/ld-aarch64/gc-relocs-257.s: Put R_AARCH64_ABS32,
> R_AARCH64_ABS16 data into data section.
>
> bfd/ChangeLog:
>
> 2017-10-18 Renlin Li <renlin.li@arm.com>
>
> * elfnn-aarch64.c (elfNN_aarch64_check_relocs): Disallow
> BFD_RELOC_AARCH64_16 in shared object const section. Disallow
> BFD_RELOC_AARCH64_32 in shared object const section under LP64.
>
>
>