[linux-next:master] BUILD REGRESSION 543b9b2fe10b2e12b8d773af65314d322f91e479
Ard Biesheuvel
ardb at kernel.org
Wed Jan 11 10:26:52 UTC 2023
On Mon, 9 Jan 2023 at 15:38, Ard Biesheuvel <ardb at kernel.org> wrote:
>
> On Mon, 9 Jan 2023 at 14:31, kernel test robot <lkp at intel.com> wrote:
> >
> > tree/branch: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master
> > branch HEAD: 543b9b2fe10b2e12b8d773af65314d322f91e479 Add linux-next specific files for 20230109
> >
> > Error/Warning: (recently discovered and may have been fixed)
> >
> > aarch64-linux-ld: ID map text too big or misaligned
>
> This is caused by linker veneers ('stubs') that are emitted into the
> ID map text section:
>
> *(.idmap.text)
> .idmap.text 0xffffffc018c04000 0x38 arch/arm64/kernel/hyp-stub.o
> .idmap.text 0xffffffc018c04038 0x28 arch/arm64/kernel/sleep.o
> 0xffffffc018c04038 cpu_resume
> .idmap.text 0xffffffc018c04060 0x38 arch/arm64/kernel/cpu-reset.o
> 0xffffffc018c04060 cpu_soft_restart
> .idmap.text 0xffffffc018c04098 0x468 arch/arm64/kernel/head.o
> 0xffffffc018c04098 init_kernel_el
> 0xffffffc018c04290 secondary_holding_pen
> 0xffffffc018c042c8 secondary_entry
> 0xffffffc018c043c4 __enable_mmu
> 0xffffffc018c0440c __cpu_secondary_check52bitva
> .idmap.text 0xffffffc018c04500 0x3c4 arch/arm64/mm/proc.o
> 0xffffffc018c04500 cpu_do_resume
> 0xffffffc018c045a8 idmap_cpu_replace_ttbr1
> 0xffffffc018c045d4 idmap_kpti_install_ng_mappings
> 0xffffffc018c047bc __cpu_setup
> *fill* 0xffffffc018c048c4 0x4
> .idmap.text.stub
> 0xffffffc018c048c8 0xf7c8 linker stubs
> 0xffffffc018c1c8c8 __idmap_text_end = .
>
> It seems unlikely that the entire .idmap.text.stub section is
> dedicated to veneers for code in .idmap.text, and the linker probably
> dumps them here because the next output section is non-executable.
>
I bisected this to
commit 5e5ff73c2e5863f93fc5fd78d178cd8f2af12464
Author: Sai Prakash Ranjan <quic_saipraka at quicinc.com>
Date: Mon Oct 17 20:04:50 2022 +0530
asm-generic/io: Add _RET_IP_ to MMIO trace for more accurate debug info
which seems entirely unrelated, but looks like it may be causing the
number of direct calls (and therefore the number of trampolines) to
increase, causing the ID map to blow up like it does.
More information about the dri-devel
mailing list