Re: [RESEND PATCH v2 0/2] efi/libstub: Ensure instruction cache coherency after kernel relocation
Huacai Chen <[email protected]>
| Newsgroups | org.kernel.vger.linux-efi,dev.linux.lists.loongarch,org.kernel.vger.linux-kernel |
|---|---|
| Message-ID | <CAAhV-H5w=UmtZt_JCR=BvZt-+F9usnmjh2dMBLHzeiE-7xBddQ@mail.gmail.com> |
On Tue, Apr 28, 2026 at 7:07 PM Ard Biesheuvel <[email protected]> wrote: > > > > On Mon, 27 Apr 2026, at 18:11, Ard Biesheuvel wrote: > > On Mon, 27 Apr 2026, at 10:47, WANG Rui wrote: > >> Changes since [v1]: > >> * Move efi_cache_sync_image() after check_platform_features(), matching arm64. > >> * Add "()" after function name in the subject. > >> > >> The EFI stub relocates the kernel image by copying it to a new memory > >> location using memcpy(). On architectures where the instruction and data > >> caches are not automatically coherent, the copied instructions may not be > >> visible to the instruction fetch unit without explicit cache maintenance. > >> > >> This series adds the missing call to efi_cache_sync_image() in the generic > >> relocation path, and provides a LoongArch implementation using the ibar > >> instruction to ensure instruction fetches observe the updated contents. > >> > >> [v1]: https://lore.kernel.org/loongarch/[email protected] > >> > >> WANG Rui (2): > >> efi/loongarch: Implement efi_cache_sync_image > >> efi/libstub: Synchronize instruction cache after kernel relocation > >> > > > > Queued up in efi/urgent - thanks. Maybe this series should cc stable? Huacai > > > > This breaks the ARM build. > > Since LoongArch is the only remaining user of efi_relocate_kernel(), I'll move it into drivers/firmware/efi/libstub/loongarch-stub.c first, and fix up these patches accordingly afterwards (no need to resend anything) >