This is the mail archive of the
binutils@sourceware.org
mailing list for the binutils project.
Re: Make gas/mips/mips.exp ELF-only
- From: "Maciej W. Rozycki" <macro at codesourcery dot com>
- To: Richard Sandiford <rdsandiford at googlemail dot com>
- Cc: <binutils at sourceware dot org>
- Date: Tue, 25 Jun 2013 19:57:56 +0100
- Subject: Re: Make gas/mips/mips.exp ELF-only
- References: <87sj0a3wxd dot fsf at talisman dot default>
On Sat, 22 Jun 2013, Richard Sandiford wrote:
> This patch removes the non-ELF support from the MIPS GAS testsuite.
>
> *-openbsd was still being classed as aout, even though the modern port
> is ELF. *-freebsd and *-kfreebsd-gnu weren't being handled at all,
> so most of the testsuite was skipped. All three, along with *-netbsd,
> are tradmips targets, so tmips should be "t" for them. This fixes
> two MIPS16 reloc failures on NetBSD.
>
> Tested on the same targets as before and applied.
This caused a new failure on mips-sgi-irix5, where I think microMIPS
tests were not previously run:
PASS: virt instructions (octeonp)
../as-new -64 -mvirt -march=mips64 -mmicromips -o dump.o .../test/binutils-src/gas/testsuite/gas/mips/virt64.s
Executing on host: sh -c {../as-new -64 -mvirt -march=mips64 -mmicromips -o dump.o .../test/binutils-src/gas/testsuite/gas/mips/virt64.s 2>&1} /dev/null gas.out (timeout = 300)
Assembler messages:
Fatal error: No compiled in support for 64 bit object file format
Assembler messages:
Fatal error: No compiled in support for 64 bit object file format
.../test/mips-sgi-irix5/gas/testsuite/../../binutils/objdump -dr --prefix-addresses --show-raw-insn -Mvirt,cp0-names=mips64r2 dump.o
Executing on host: sh -c {.../test/mips-sgi-irix5/gas/testsuite/../../binutils/objdump -dr --prefix-addresses --show-raw-insn -Mvirt,cp0-names=mips64r2 dump.o >dump.out 2>gas.stderr} /dev/null (timeout = 300)
regexp_diff match failure
regexp "^[0-9a-f]+ <[^>]*> 587d 00e7 dmfgc0 v1,c0_taghi$"
line "00000000 <foo> 4063e800 mfgc0 v1,c0_taghi"
regexp_diff match failure
regexp "^[0-9a-f]+ <[^>]*> 5974 28e7 dmfgc0 a7,\$20,5$"
line "00000004 <foo+0x4> 406ba005 mfgc0 t3,$20,5"
regexp_diff match failure
regexp "^[0-9a-f]+ <[^>]*> 5ae2 02e7 dmtgc0 s7,c0_entrylo0$"
line "00000008 <foo+0x8> 40771200 mtgc0 s7,c0_entrylo0"
regexp_diff match failure
regexp "^[0-9a-f]+ <[^>]*> 58ee 12e7 dmtgc0 a3,\$14,2$"
line "0000000c <foo+0xc> 40677202 mtgc0 a3,$14,2"
regexp_diff match failure
regexp "^ \.\.\.$"
line "00000010 <foo+0x10> 42000028 hypcall"
extra lines in dump.out starting with "^00000014 <foo+0x14> 4212b028 hypcall 0x256$"
EOF from .../test/binutils-src/gas/testsuite/gas/mips/micromips@virt64.d
FAIL: virt64 instructions (micromips)
-- and as you can also see the test harness proceeds to running `objdump'
on a binary produced by a preceding test, risking a false positive. I
think the harness should bail out right away if assembly failed. I'm not
sure if this is a regression or if the harness has always worked like it
does.
Maciej