This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: New GAS testsuite failures after your THUMB32 patch
- From: Zack Weinberg <zack at codesourcery dot com>
- To: Nick Clifton <nickc at redhat dot com>
- Cc: binutils at sources dot redhat dot com, jbeulich at novell dot com
- Date: Tue, 24 May 2005 11:32:07 -0700
- Subject: Re: New GAS testsuite failures after your THUMB32 patch
- References: <m3is1akv9e.fsf@redhat.com>
On Mon, 2005-05-23 at 12:59 +0100, Nick Clifton wrote:
> Hi Zack,
>
> I am seeing some new gas testsuite failures after the Thumb32 patch
> was committed. Failures such as:
>
> FAIL: ARM Thumb-compat pseudos
> FAIL: Thumb ARM-compat pseudos
> FAIL: VFP errors
> FAIL: .req errors
> FAIL: Invalid use of r15 errors
> FAIL: Valid ARM, invalid Thumb
> FAIL: iWMMXt errors
> FAIL: Thumb instructions
> FAIL: 32-bit Thumb instructions
> FAIL: TLS
With a CVS checkout from this morning, I don't see all of the failures
you list, but I do see some. Here's my complete list of FAILs with
cross-binutils to all the targets you're trying. (I'm pretty sure the
ld failures are unrelated, but I'm listing them just in case.) In case
it matters, what host are you using? I'm on x86_64-linux.
arm-aout: === ld tests ===
arm-aout:FAIL: ld-scripts/data
arm-aout:FAIL: ld-scripts/provide-1
arm-aout:FAIL: ld-scripts/provide-2
arm-aout:FAIL: ld-scripts/size-1
arm-epoc-pe: === ld tests ===
arm-epoc-pe:FAIL: align1
arm-none-eabi: === gas tests ===
arm-none-eabi:FAIL: 32-bit Thumb instructions
arm-none-eabi:FAIL: group section
arm-none-eabi:FAIL: group section with multiple sections of same name
arm-nto: === gas tests ===
arm-nto:FAIL: 32-bit Thumb instructions
arm-symbianelf: === gas tests ===
arm-symbianelf:FAIL: 32-bit Thumb instructions
arm-symbianelf:FAIL: group section
arm-symbianelf:FAIL: group section with multiple sections of same name
arm-vxworks: === gas tests ===
arm-vxworks:FAIL: TLS
arm-vxworks:FAIL: Thumb instructions
arm-vxworks:FAIL: 32-bit Thumb instructions
arm-wince-pe: === gas tests ===
arm-wince-pe:FAIL: ARM basic instructions (WinCE version)
arm-wince-pe:FAIL: ARM arm7t (WinCE version)
arm-wince-pe:FAIL: ARM architecture 4t instructions
arm-wince-pe:FAIL: immediate expressions
arm-wince-pe:FAIL: OFFSET_IMM regression
arm-wince-pe: === ld tests ===
arm-wince-pe:FAIL: align1
arm-wince-pe:FAIL: ld-scripts/align2a
arm-wince-pe:FAIL: ld-scripts/align2b
arm-wince-pe:FAIL: ld-scripts/data
arm-wince-pe:FAIL: ld-scripts/defined2
arm-wince-pe:FAIL: ld-scripts/defined3
arm-wince-pe:FAIL: ld-scripts/provide-1
arm-wince-pe:FAIL: ld-scripts/provide-2
arm-wince-pe:FAIL: ld-scripts/size-1
xscale-coff: === gas tests ===
xscale-coff:FAIL: Thumb instructions
xscale-coff:FAIL: 32-bit Thumb instructions
xscale-elf: === gas tests ===
xscale-elf:FAIL: 32-bit Thumb instructions
xscale-elf:FAIL: group section
xscale-elf:FAIL: group section with multiple sections of same name
I will investigate all of the gas failures further. On a cursory
inspection, none of them seems to have anything to do with Jan's macro
changes; they all appear to be either off-by-8 errors in relocs, or
places where tc-arm.c is expecting to be run on a 32-bit host.
zw