[RFC] RISC-V: Validate ALIGN relocation padding bounds
Dorian Vail
snozns@outlook.com
Mon Nov 16 02:16:44 GMT 2026
Hello Binutils maintainers,
I would appreciate feedback on a proposed validation change for RISC-V ALIGN relaxation. The linker should check that the complete padding span fits within the input section before performing alignment arithmetic or modifying section contents. Invalid spans should fail the link with a diagnostic.
I compared stock and historical test builds with assembler-generated alignment cases. The valid objects linked and preserved alignment with compressed instructions enabled and disabled. A master-based patch and its regression results are still being prepared.
Bugzilla #34844 concerns non-termination in the ALIGN path. Could you advise whether the padding validation should be reviewed together with the work on that issue?
Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://sourceware.org/pipermail/binutils/attachments/20261116/691472f4/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: F7-binutils-2.47-local-only.patch
Type: application/octet-stream
Size: 1470 bytes
Desc: F7-binutils-2.47-local-only.patch
URL: <https://sourceware.org/pipermail/binutils/attachments/20261116/691472f4/attachment.obj>
More information about the Binutils
mailing list