Re: [PATCH v2 3/3] btf_loader.c: Infer alignment info

[Arnaldo Carvalho de Melo <acme@kernel.org>]

Em Thu, Oct 21, 2021 at 09:31:36PM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Mon, Oct 18, 2021 at 02:16:21PM +0100, Douglas RAILLARD escreveu:
> > From: Douglas Raillard <douglas.raillard@arm.com>
> > 
> > BTF does not carry alignment information, but it carries the offset in
> > structs. This allows inferring the original alignment, yielding a C
> > header dump that is not identical to the original C code, but is
> > guaranteed to lead to the same memory layout.
> > 
> > This allows using the output of pahole in another program to poke at
> > memory, with the assurance that we will not read garbage.
> > 
> > Note: Since the alignment is inferred from the offset, it sometimes
> > happens that the offset was already correctly aligned, which means the
> > inferred alignment will be smaller than in the original source. This
> > does not impact the ability to read existing structs, but it could
> > impact creating such struct if other client code expects higher
> > alignment than the one exposed in the generated header.
> 
> this one makes btfdiff fail, example:
> 
> 
> @@ -125578,7 +125640,7 @@ struct xt_entry_match {
>  			struct xt_match * match;         /*     8     8 */
>  		} kernel;                                /*     0    16 */
>  		__u16              match_size;           /*     0     2 */
> -	} u;                                             /*     0    32 */
> +	} u;            /*     0    32 */
>  	unsigned char              data[];               /*    32     0 */
> 
>  	/* size: 32, cachelines: 1, members: 2 */
> 
> Why the change in the generated source code comment alignment?

Since this is inferred and DWARF has the alignment info explicitely, we
can't really compare, so I'll stick this to btfdiff.

diff --git a/btfdiff b/btfdiff
index 77543630d1965b5e..cbdf65285cd90f62 100755
--- a/btfdiff
+++ b/btfdiff
@@ -33,6 +33,7 @@ ${pahole_bin} -F dwarf \
 	      --show_private_classes $dwarf_input > $dwarf_output
 ${pahole_bin} -F btf \
 	      --sort \
+	      --suppress_aligned_attribute \
 	      --suppress_packed \
 	      $btf_input > $btf_output
 
 
> There are some other differences, I'll check tomorrow.

I got sidetracked, will try to reduce the differences somehow, but
probably as patches on top of yours, to test this the best is to use
fullcircle from BTF info, which I'll also look into as I think you
reported issues with pfunct's --compile option.

- Arnaldo