[Bug gas/32435] [2.44 Regression] gas produces unaligned sections on arm-linux-gnueabihf for kernel build

[jbeulich at suse dot com]

https://sourceware.org/bugzilla/show_bug.cgi?id=32435

--- Comment #11 from Jan Beulich <jbeulich at suse dot com> ---
(In reply to Eric Botcazou from comment #10)
> > It doesn't need to be Arm32-specific ones. Such a .rodata -> .rodata.str.*
> > conversion looks pretty generic. Just that I'm unaware of anything along
> > these lines being done in the upstream gas.
> 
> Yes, that change is done in the compiler.

Except that it was an assembly file which was supplied, with no use of
.rodata.str.*.

>  But, in the testcase, the .rodata
> section itself is misaligned:
> 
> Sections:
> Idx Name          Size      VMA       LMA       File off  Algn
>   0 .text         00002454  00000000  00000000  00000034  2**2
>                   CONTENTS, ALLOC, LOAD, RELOC, READONLY, CODE
>   1 .data         00000101  00000000  00000000  00002488  2**2
>                   CONTENTS, ALLOC, LOAD, RELOC, DATA
>   2 .bss          00000000  00000000  00000000  00002589  2**2
>                   ALLOC
>   3 .init.text    00000038  00000000  00000000  00002589  2**2
>                   CONTENTS, ALLOC, LOAD, RELOC, READONLY, CODE
>   4 .rodata       0000104c  00000000  00000000  000025c1  2**2
>                   CONTENTS, ALLOC, LOAD, RELOC, READONLY, DATA

I'm afraid there's no concept of "aligning" within the (object) file. Yet still
I was thinking of exactly such (generally speaking wrong) assumptions, making
me add the post-commit-message RFC remark in the original submission of the
patch (see https://sourceware.org/pipermail/binutils/2024-October/137085.html).

-- 
You are receiving this mail because:
You are on the CC list for the bug.