Re: [PATCH] MIPS: fix readelf -S bintest test for N64 triples

["Maciej W. Rozycki" <macro@orcam.me.uk>]

On Thu, 17 Aug 2023, YunQiang Su wrote:

> MIPS N64 has different section layout. Let's add a new file
> readelf.s-64-tmips.
> 
> It can fix this test fail on mips64*-linux-gnuabi64 and mips64*-openbsd.

 I actually had to chase readelf.exp and dive into it so as to find out 
whether and why this approach is correct.  This could have been mentioned 
in the description.

 Also due to how the test is called it wasn't clear to me from the 
description itself what test is actually affected.

> diff --git a/binutils/testsuite/binutils-all/readelf.s-64-tmips b/binutils/testsuite/binutils-all/readelf.s-64-tmips
> new file mode 100644
> index 00000000000..5c19027d0b7
> --- /dev/null
> +++ b/binutils/testsuite/binutils-all/readelf.s-64-tmips
> @@ -0,0 +1,25 @@
> +There are .* section headers, starting at offset .*:
> +
> +Section Headers:
> + +\[Nr\] Name +Type +Address +Offset
> + +Size +EntSize +Flags +Link +Info +Align
> + +\[ 0\] +NULL +0000000000000000 +00000000
> + +0000000000000000 +0000000000000000 +0 +0 +0
> + +\[ 1\] .text +PROGBITS +0000000000000000 +00000040
> + +00000000000000.. +0000000000000000 +AX +0 +0 +.*
> + +\[ 2\] .rel.+text +REL. +0+ +0+.*
> + +000000000000001. +000000000000001. +I +. +1 +8
> + +\[ 3\] .data +PROGBITS +0000000000000000 +000000(48|50)
> + +0000000000000010 +0000000000000000 +WA +0 +0 +.*
> + +\[ 4\] .bss +NOBITS +0000000000000000 +00000060
> + +0000000000000000 +0000000000000000 +WA +0 +0 +.*
> +# .MIPS.options, .MIPS.abiflags, .pdr, .gnu.attributes here
> +#...
> + +\[ .\] .symtab +SYMTAB +0000000000000000 +0+.*
> + +0+.* +0000000000000018 +10 +10 +8
> + +\[10\] .strtab +STRTAB +0000000000000000 +0+.*
> + +0+.* +0000000000000000 .* +0 +0 +1
> + +\[11\] .shstrtab +STRTAB +0000000000000000 +[0-9a-f]+
> + +00000000000000.. +0000000000000000 .* +0 +0 +.*
> +Key to Flags:
> +#...

 Since the scope for this dump is so narrow there is no need to 
wildcard-match output produced.  It can be an exact match for easier 
detection of unwanted changes in output, just as the original readelf.s 
dump used to be.

 Also #... at the end is unusual, #pass can be used to terminate matching 
successfully right away rather than wading through the rest of output only 
to ignore it.

 I have committed the change with these updates made for you, posted 
separately, thank you for your contribution.

  Maciej