From: "Richard Earnshaw (lists)" <Richard.Earnshaw@arm.com>
To: Bernd Edlinger <bernd.edlinger@hotmail.de>,
Richard Biener <rguenther@suse.de>
Cc: "gcc-patches@gcc.gnu.org" <gcc-patches@gcc.gnu.org>,
Ramana Radhakrishnan <ramana.radhakrishnan@arm.com>,
Kyrill Tkachov <kyrylo.tkachov@foss.arm.com>,
Eric Botcazou <ebotcazou@adacore.com>, Jeff Law <law@redhat.com>,
Jakub Jelinek <jakub@redhat.com>
Subject: Re: [PATCHv5] Fix not 8-byte aligned ldrd/strd on ARMv5 (PR 89544)
Date: Thu, 05 Sep 2019 09:21:00 -0000 [thread overview]
Message-ID: <d42db60d-44cf-c167-a26e-c226e3197c6d@arm.com> (raw)
In-Reply-To: <f49000af-3c18-efcc-043c-481be60afe9b@arm.com>
On 04/09/2019 16:48, Richard Earnshaw (lists) wrote:
> On 04/09/2019 16:00, Bernd Edlinger wrote:
>> On 9/4/19 4:14 PM, Richard Earnshaw (lists) wrote:
>>> On 04/09/2019 14:28, Bernd Edlinger wrote:
>>>> On 9/4/19 2:53 PM, Richard Earnshaw (lists) wrote:
>>>>> Index: gcc/testsuite/gcc.target/arm/unaligned-argument-2.c
>>>>> ===================================================================
>>>>> --- gcc/testsuite/gcc.target/arm/unaligned-argument-2.c
>>>>> (Revision 0)
>>>>> +++ gcc/testsuite/gcc.target/arm/unaligned-argument-2.c
>>>>> (Arbeitskopie)
>>>>> @@ -0,0 +1,19 @@
>>>>> +/* { dg-do compile } */
>>>>> +/* { dg-require-effective-target arm_arm_ok } */
>>>>> +/* { dg-require-effective-target arm_ldrd_strd_ok } */
>>>>> +/* { dg-options "-marm -mno-unaligned-access -O3" } */
>>>>> +
>>>>> +struct s {
>>>>> +Â int a, b;
>>>>> +} __attribute__((aligned(8)));
>>>>> +
>>>>> +struct s f0;
>>>>> +
>>>>> +void f(int a, int b, int c, int d, int e, struct s f)
>>>>> +{
>>>>> +Â f0 = f;
>>>>> +}
>>>>> +
>>>>> +/* { dg-final { scan-assembler-times "ldrd" 0 } } */
>>>>> +/* { dg-final { scan-assembler-times "strd" 0 } } */
>>>>> +/* { dg-final { scan-assembler-times "stm" 1 } } */
>>>>>
>>>>> I don't think this test is right. While we can't use an LDRD to
>>>>> load the argument off the stack, there's nothing wrong with using
>>>>> an STRD to then store the value to f0 (as that is 8-byte aligned).
>>>>> So the second and third scan-assembler tests are meaningless.
>>>>>
>>>>
>>>> Ah, that is very similar to the unaligned-memcpy-2/3.c,
>>>> see https://gcc.gnu.org/ml/gcc-patches/2019-09/msg00157.html
>>>>
>>>> initially that is a movdi,
>>>> then in subreg1 it is split in two movsi
>>>> which is then re-assembled as ldm
>>>>
>>>>
>>>> Not sure if that is intended in that way.
>>>>
>>>>
>>>
>>> Yeah, these are causing me some problems too, but that's because with
>>> some changes I'm working on I now see the compiler using r4 and r5,
>>> which leads to prologue and epilogue stores that distort the results.
>>>
>>> Tests like this are generally fragile - I hate 'em!!!!
>>>
>>
>> Yeah, that changed since r275063 introduced the unaligned-load/storedi
>>
>> r275063 | edlinger | 2019-08-30 12:38:37 +0200 (Fr, 30. Aug 2019) | 10
>> Zeilen
>> Geänderte Pfade:
>> Â Â Â M /trunk/gcc/ChangeLog
>> Â Â Â M /trunk/gcc/config/arm/arm.c
>> Â Â Â M /trunk/gcc/config/arm/arm.md
>> Â Â Â M /trunk/gcc/config/arm/neon.md
>>
>> 2019-08-30 Bernd Edlinger <bernd.edlinger@hotmail.de>
>>
>> Â Â Â Â Â Â Â Â * config/arm/arm.md (unaligned_loaddi,
>> Â Â Â Â Â Â Â Â unaligned_storedi): New unspec insn patterns.
>> Â Â Â Â Â Â Â Â * config/arm/neon.md (unaligned_storev8qi): Likewise.
>> Â Â Â Â Â Â Â Â * config/arm/arm.c (gen_cpymem_ldrd_strd): Use unaligned_loaddi
>> Â Â Â Â Â Â Â Â and unaligned_storedi for 4-byte aligned memory.
>> Â Â Â Â Â Â Â Â (arm_block_set_aligned_vect): Use unaligned_storev8qi for
>> Â Â Â Â Â Â Â Â 4-byte aligned memory.
>>
>> Since other than the movdi they are not split up but stay as ldrd/strd.
>> But for some unknown reason ira assigns r4-5 to those although also
>> r1-2 would be available. :-(
>>
>
> r1-r2 can't be used in Arm state as the register has to start on an even
> boundary. But ira has already used r3 for the address of the store (it
> could have picked r1) and now r4-r5 is the next even-numbered pair. So
> we end up with needing to save some call-clobbered regs.
>
> R.
>>
>> Bernd.
>>
>
One possible trick to stabilize the test is to insert an asm that
clobbers r4 and r5 and forces the prologue/epilogue code to always save
and restore them. Then we can account for those prologue/epilogue
consistently (at least, modulo the arm_prefer_ldrd_strd condition).
R.
next prev parent reply other threads:[~2019-09-05 9:21 UTC|newest]
Thread overview: 50+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-03-10 12:51 [PATCHv2] " Bernd Edlinger
2019-03-19 14:01 ` [PING] " Bernd Edlinger
2019-03-21 11:26 ` Richard Biener
2019-03-22 17:47 ` Bernd Edlinger
2019-03-25 9:28 ` Richard Biener
2019-07-30 22:13 ` [PATCHv3] " Bernd Edlinger
2019-07-31 13:17 ` Richard Earnshaw (lists)
2019-08-01 11:19 ` Bernd Edlinger
2019-08-02 9:10 ` Richard Earnshaw (lists)
2019-08-02 13:11 ` Richard Biener
2019-08-02 19:01 ` Bernd Edlinger
2019-08-08 14:20 ` [PATCHv4] " Bernd Edlinger
2019-08-14 10:54 ` [PING] " Bernd Edlinger
2019-08-14 12:27 ` Richard Biener
2019-08-14 22:26 ` Bernd Edlinger
2019-08-15 8:58 ` Richard Biener
2019-08-15 12:38 ` Bernd Edlinger
2019-08-15 13:03 ` Richard Biener
2019-08-15 14:33 ` Richard Biener
2019-08-15 15:28 ` Bernd Edlinger
2019-08-15 17:42 ` Richard Biener
2019-08-15 21:19 ` [PATCHv5] " Bernd Edlinger
2019-08-20 5:38 ` Jeff Law
2019-08-20 15:04 ` John David Anglin
[not found] ` <0d39b64f-67d9-7857-cf4e-36f09c0dc15e@bell.net>
2019-08-20 16:03 ` Fwd: " Bernd Edlinger
2019-09-04 12:53 ` Richard Earnshaw (lists)
2019-09-04 13:29 ` Bernd Edlinger
2019-09-04 14:14 ` Richard Earnshaw (lists)
2019-09-04 15:00 ` Bernd Edlinger
2019-09-04 15:48 ` Richard Earnshaw (lists)
2019-09-05 9:21 ` Richard Earnshaw (lists) [this message]
2019-09-05 9:35 ` Bernd Edlinger
2019-09-06 10:15 ` Bernd Edlinger
2019-09-06 10:18 ` Richard Earnshaw (lists)
2019-08-15 21:27 ` [PATCH] Sanitizing the middle-end interface to the back-end for strict alignment Bernd Edlinger
2019-08-17 10:11 ` Bernd Edlinger
2019-08-23 0:01 ` Jeff Law
2019-08-23 0:05 ` Jeff Law
2019-08-23 15:15 ` [PING] " Bernd Edlinger
2019-08-27 10:07 ` Kyrill Tkachov
2019-08-28 11:50 ` Bernd Edlinger
2019-08-28 12:01 ` Kyrill Tkachov
2019-08-28 13:54 ` Christophe Lyon
2019-08-28 21:48 ` Bernd Edlinger
2019-08-29 9:09 ` Kyrill Tkachov
2019-08-29 10:00 ` Christophe Lyon
2019-08-29 22:57 ` Bernd Edlinger
2019-08-30 10:07 ` Kyrill Tkachov
2019-08-30 15:22 ` Christophe Lyon
2019-08-14 11:56 ` [PATCHv3] Fix not 8-byte aligned ldrd/strd on ARMv5 (PR 89544) Richard Biener
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=d42db60d-44cf-c167-a26e-c226e3197c6d@arm.com \
--to=richard.earnshaw@arm.com \
--cc=bernd.edlinger@hotmail.de \
--cc=ebotcazou@adacore.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=jakub@redhat.com \
--cc=kyrylo.tkachov@foss.arm.com \
--cc=law@redhat.com \
--cc=ramana.radhakrishnan@arm.com \
--cc=rguenther@suse.de \
/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).