public inbox for binutils@sourceware.org
 help / color / mirror / Atom feed
From: Tristan Gingold <gingold@adacore.com>
To: Matthias Klose <doko@ubuntu.com>
Cc: Andreas Krebbel <krebbel@linux.vnet.ibm.com>,
	binutils@sources.redhat.com, schwidefsky@de.ibm.com
Subject: Re: [PATCH] S/390: Make as to accept all instructions if no -march option given
Date: Wed, 23 Mar 2011 16:02:00 -0000	[thread overview]
Message-ID: <AD319EDC-A256-4098-B4BB-E37A9A5F2A2C@adacore.com> (raw)
In-Reply-To: <4D891DB6.40603@ubuntu.com>


On Mar 22, 2011, at 11:07 PM, Matthias Klose wrote:

> Could this be applied to the 2.21 branch? Doesn't accept all instructions if no
> -march option given and breaks the glibc build if -march is not explicitly
> specified (Debian #618751).

Done.

> 
> Thank, Matthias
> 
> On 24.11.2010 17:39, Andreas Krebbel wrote:
>> Hi,
>> 
>> for S/390 currently mimics the GCC behaviour when no -march switch is
>> specified.  This means that it defaults to -march=g5 for -m31 and to
>> -march=z900 for -m64.  Due to that the vers1 testcase currently fails
>> when building GCC with a different default -march argument since it
>> relies on an "as" execution without -march accepting all valid
>> instructions.
>> 
>> The attached patch fixes this by defaulting to the highest CPU and the
>> most generic mode (z/Architecture mode).  For S/390 this is the most
>> generic instruction set containing all possible instructions.
>> 
>> Tested on s390 and s390x. The vers1 testcase is fixed with that change.
>> 
>> I'll commit the patch tomorrow if nobody objects.
>> 
>> Bye,
>> 
>> -Andreas-
>> 
>> 
>> 2010-11-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
>> 
>> 	* config/tc-s390.c (current_cpu): Initialize with latest CPU.
>> 	(init_default_arch): Default to z/Architecture mode if CPU provides it.
>> 	Remove the check setting the CPU default.
>> 
>> 	
>> 2010-11-24  Andreas Krebbel  <Andreas.Krebbel@de.ibm.com>
>> 
>> 	* opcode/s390.h (enum s390_opcode_cpu_val): Add S390_OPCODE_MAXCPU.
>> 
>> 
>> Index: gas/config/tc-s390.c
>> ===================================================================
>> --- gas/config/tc-s390.c.orig
>> +++ gas/config/tc-s390.c
>> @@ -38,8 +38,11 @@ static char *default_arch = DEFAULT_ARCH
>> /* Either 32 or 64, selects file format.  */
>> static int s390_arch_size = 0;
>> 
>> +/* If no -march option was given default to the highest available CPU.
>> +   Since with S/390 a newer CPU always supports everything from its
>> +   predecessors this will accept every valid asm input.  */
>> +static unsigned int current_cpu = S390_OPCODE_MAXCPU - 1;
>> static unsigned int current_mode_mask = 0;
>> -static unsigned int current_cpu = -1U;
>> 
>> /* Whether to use user friendly register names. Default is TRUE.  */
>> #ifndef TARGET_REG_NAMES_P
>> @@ -328,18 +331,12 @@ init_default_arch (void)
>> 
>>   if (current_mode_mask == 0)
>>     {
>> -      if (s390_arch_size == 32)
>> +      /* Default to z/Architecture mode if the CPU supports it.  */
>> +      if (current_cpu < S390_OPCODE_Z900)
>> 	current_mode_mask = 1 << S390_OPCODE_ESA;
>>       else
>> 	current_mode_mask = 1 << S390_OPCODE_ZARCH;
>>     }
>> -  if (current_cpu == -1U)
>> -    {
>> -      if (current_mode_mask == (1 << S390_OPCODE_ESA))
>> -	current_cpu = S390_OPCODE_G5;
>> -      else
>> -	current_cpu = S390_OPCODE_Z900;
>> -    }
>> }
>> 
>> /* Called by TARGET_FORMAT.  */
>> Index: include/opcode/s390.h
>> ===================================================================
>> --- include/opcode/s390.h.orig
>> +++ include/opcode/s390.h
>> @@ -39,7 +39,8 @@ enum s390_opcode_cpu_val
>>     S390_OPCODE_Z9_109,
>>     S390_OPCODE_Z9_EC,
>>     S390_OPCODE_Z10,
>> -    S390_OPCODE_Z196
>> +    S390_OPCODE_Z196,
>> +    S390_OPCODE_MAXCPU
>>   };
>> 
>> /* The opcode table is an array of struct s390_opcode.  */
> 

      reply	other threads:[~2011-03-23 16:02 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-11-24 16:39 Andreas Krebbel
2011-03-22 22:08 ` Matthias Klose
2011-03-23 16:02   ` Tristan Gingold [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=AD319EDC-A256-4098-B4BB-E37A9A5F2A2C@adacore.com \
    --to=gingold@adacore.com \
    --cc=binutils@sources.redhat.com \
    --cc=doko@ubuntu.com \
    --cc=krebbel@linux.vnet.ibm.com \
    --cc=schwidefsky@de.ibm.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).