From: Joey Ye <joey.ye.cc@gmail.com>
To: "Bin.Cheng" <amker.cheng@gmail.com>
Cc: gcc-patches <gcc-patches@gcc.gnu.org>, richard.sandiford@arm.com
Subject: Re: [PATCH AArch64]Fix test failure for pr84682-2.c
Date: Thu, 30 Aug 2018 10:21:00 -0000 [thread overview]
Message-ID: <CAL0py25m3UVXz6VH4wCA+3EZjp+2+uasxKH=8z+=Ugfg5onMTA@mail.gmail.com> (raw)
In-Reply-To: <CAHFci2-NkyLa95FwropaX2x+GqXns3mnp12fFJP5_KD_RiBKgA@mail.gmail.com>
Hi Bin & Richard,
It is not as simple as keeping the assertion, which still fails even
with the change in reorg.c. The testing result is as following:
I. With Bin's patch version 2 (removing the assertion in aarch64.c and
adding the check in reorg.c): pr84682-2.c passes
II. With Richard's suggestion to keep the assertion in aarch64, but
adding the check in reorg.c: pr84682-2.c fails
Apparently there is a different path that reaches the assertion.
With II:
/build/trunk/src/gcc/gcc/testsuite/gcc.dg/torture/pr84682-2.c: In
function 'b': /build/trunk/src/gcc/gcc/testsuite/gcc.dg/torture/pr84682-2.c:10:1:
internal compiler error: in aarch64_classify_address, at
config/aarch64/aarch64.c:5721
0xfa4071 aarch64_classify_address
/build/trunk/src/gcc/gcc/config/aarch64/aarch64.c:5720
0xfa94f3 aarch64_legitimate_address_hook_p
/build/trunk/src/gcc/gcc/config/aarch64/aarch64.c:6003
0xb85661 strict_memory_address_addr_space_p(machine_mode, rtx_def*,
unsigned char)
/build/trunk/src/gcc/gcc/reload.c:2177
0xb75da9 constrain_operands(int, unsigned long)
/build/trunk/src/gcc/gcc/recog.c:2706
0xb761a0 extract_constrain_insn(rtx_insn*)
/build/trunk/src/gcc/gcc/recog.c:2210
0xa6dd57 check_rtl
/build/trunk/src/gcc/gcc/lra.c:2182
0xa737db lra(_IO_FILE*)
/build/trunk/src/gcc/gcc/lra.c:2616
0xa25989 do_reload
/build/trunk/src/gcc/gcc/ira.c:5469
0xa25989 execute
Current trunk without any patch:
/build/trunk/src/gcc/gcc/testsuite/gcc.dg/torture/pr84682-2.c: In
function 'b': /build/trunk/src/gcc/gcc/testsuite/gcc.dg/torture/pr84682-2.c:9:3:
internal compiler error: in aarch64_classify_address, at
config/aarch64/aarch64.c:5721
0xfa4011 aarch64_classify_address
/build/trunk/src/gcc/gcc/config/aarch64/aarch64.c:5720
0xfa9493 aarch64_legitimate_address_hook_p
/build/trunk/src/gcc/gcc/config/aarch64/aarch64.c:6003
0xb74cf3 memory_address_addr_space_p(machine_mode, rtx_def*, unsigned char)
/build/trunk/src/gcc/gcc/recog.c:1334
0xb74cf3 address_operand(rtx_def*, machine_mode)
/build/trunk/src/gcc/gcc/recog.c:1073
0xb74cf3 asm_operand_ok(rtx_def*, char const*, char const**)
/build/trunk/src/gcc/gcc/recog.c:1817
0x75e591 expand_asm_stmt
/build/trunk/src/gcc/gcc/cfgexpand.c:3135
0x766d67 expand_gimple_stmt_1
/build/trunk/src/gcc/gcc/cfgexpand.c:3572
0x766d67 expand_gimple_stmt
/build/trunk/src/gcc/gcc/cfgexpand.c:3734
0x768ce7 expand_gimple_basic_block
More places need to be patched.
Thanks,
Joey
On Thu, Aug 30, 2018 at 2:02 AM Bin.Cheng <amker.cheng@gmail.com> wrote:
>
> On Thu, Aug 30, 2018 at 2:47 AM Richard Sandiford
> <richard.sandiford@arm.com> wrote:
> >
> > Joey Ye <joey.ye.cc@gmail.com> writes:
> > > diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
> > > index 07c55b1..9e965ab 100644
> > > --- a/gcc/config/aarch64/aarch64.c
> > > +++ b/gcc/config/aarch64/aarch64.c
> > > @@ -5674,9 +5674,6 @@ aarch64_classify_address (struct aarch64_address_info *info,
> > > && (code != POST_INC && code != REG))
> > > return false;
> > >
> > > - gcc_checking_assert (GET_MODE (x) == VOIDmode
> > > - || SCALAR_INT_MODE_P (GET_MODE (x)));
> > > -
> > > switch (code)
> > > {
> > > case REG:
> > > diff --git a/gcc/recog.c b/gcc/recog.c
> > > index 0a8fa2c..510aba2 100644
> > > --- a/gcc/recog.c
> > > +++ b/gcc/recog.c
> > > @@ -1070,6 +1070,11 @@ general_operand (rtx op, machine_mode mode)
> > > int
> > > address_operand (rtx op, machine_mode mode)
> > > {
> > > + /* Wrong mode for an address expr. */
> > > + if (GET_MODE (op) != VOIDmode
> > > + && ! SCALAR_INT_MODE_P (GET_MODE (op)))
> > > + return false;
> > > +
> > > return memory_address_p (mode, op);
> > > }
> > >
> >
> > The address_operand part is OK, thanks.
> >
> > I think we should keep the assert in aarch64_classify_address, since
> > IMO it's a bug for anything else to reach that point.
>
> Hi Joey,
> Could you help me update the patch as suggested by Richard and commit
> it please? My new assignment is still on the way.
> Thanks very much!
>
> Thanks,
> bin
> >
> > Richard
next prev parent reply other threads:[~2018-08-30 10:21 UTC|newest]
Thread overview: 13+ messages / expand[flat|nested] mbox.gz Atom feed top
2018-03-16 11:43 Bin Cheng
2018-03-16 11:53 ` Kyrill Tkachov
2018-03-17 9:20 ` Richard Sandiford
2018-03-22 11:11 ` Bin.Cheng
2018-04-24 15:14 ` Bin.Cheng
2018-05-16 8:37 ` Kyrill Tkachov
2018-08-29 15:50 ` Joey Ye
2018-08-29 18:47 ` Richard Sandiford
2018-08-30 1:02 ` Bin.Cheng
2018-08-30 10:21 ` Joey Ye [this message]
2018-08-30 10:28 ` Joey Ye
2018-08-30 12:24 ` Richard Sandiford
2018-12-12 11:29 ` Richard Earnshaw (lists)
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='CAL0py25m3UVXz6VH4wCA+3EZjp+2+uasxKH=8z+=Ugfg5onMTA@mail.gmail.com' \
--to=joey.ye.cc@gmail.com \
--cc=amker.cheng@gmail.com \
--cc=gcc-patches@gcc.gnu.org \
--cc=richard.sandiford@arm.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).