This is the mail archive of the binutils@sourceware.org mailing list for the binutils project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Patch] [binutils][testsuite] Allow multiple lines of "as" in testsuite.


Hi Matthew,

> Great, would the attached extra comments be OK?

In theory yes.  In practice no.  Actually it is not the extra comments
that are the problem, but the original patch.  Now that I have tested
it locally, I found that it introduces some new testsuite failures. :-(

For example:

Checking Binutils in: s390-linux ... GAS REGRESSION: s390x opcode   

Checking Binutils in: mips64-linux ... LD REGRESSION: ld-mips-elf/no-shared-1-n64    
  LD REGRESSION: MIPS ELF got reloc n64
  LD REGRESSION: MIPS ELF xgot reloc n64
  LD REGRESSION: MIPS rel32 o32  
  LD REGRESSION: MIPS rel64 n64  
  LD REGRESSION: PIC and non-PIC test 3
  LD REGRESSION: PIC and non-PIC test 4
 
(These failures also show up when testing a toolchain configured with
 mipsel-linux-gnu and also with mipsisa32el-linux-gnu).

The worst affected toolchain however is the PowerPC:

 powerpc-ibm-aix5.2.0 ... LD ERROR: powerpc-ibm-aix5.2.0-readelf does not exist 
  LD REGRESSION: --gc-sections -r without -e 
  LD ERROR: powerpc-ibm-aix5.2.0-as does not exist 
  LD ERROR: /work/sources/binutils/current/ld/testsuite/ld-libs/lib-1.s: assembly failed  
  LD ERROR: powerpc-ibm-aix5.2.0-as does not exist 
  LD ERROR: /work/sources/binutils/current/ld/testsuite/ld-libs/lib-2.s: assembly failed  
  LD ERROR: powerpc-ibm-aix5.2.0-as does not exist 
  LD ERROR: /work/sources/binutils/current/ld/testsuite/ld-misc/start.s: assembly failed  
  LD ERROR: powerpc-ibm-aix5.2.0-as does not exist 
  [...and so on, there are lots more like this...]

Please could you have a look at these.  I suspect that the S390 and MIPS failures
might be due to a variable being set by a previous test which is not cleared before
the failing test runs.  (But this is just a guess).  The PowerPC ones all appear to
be because the new code is not checking for the presence of a locally built assembler
rather than one installed into the execution path.  But again that is just off the
top of my head.

Cheers
  Nick


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]