public inbox for gcc-bugs@sourceware.org
help / color / mirror / Atom feed
From: "anemo at mba dot ocn.ne.jp" <gcc-bugzilla@gcc.gnu.org>
To: gcc-bugs@gcc.gnu.org
Subject: [Bug c/45819] New: [4.5 Regression] unexpected unaligned access to volatile int
Date: Tue, 28 Sep 2010 17:51:00 -0000	[thread overview]
Message-ID: <bug-45819-4@http.gcc.gnu.org/bugzilla/> (raw)

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=45819

           Summary: [4.5 Regression] unexpected unaligned access to
                    volatile int
           Product: gcc
           Version: 4.5.1
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: anemo@mba.ocn.ne.jp


An unexpected unaligned access instructions for volatile int are
generated on gcc 4.5 or 4.6 with this test case.

struct st {
    int ptr;
} __attribute__ ((packed));

int foo(struct st *st)
{
    int v = *(volatile int *)&st->ptr;
    return v;
}

This test case is derived from something like this in ARM linux kernel.

struct st {
    unsigned int ptr;
} __attribute__ ((packed));
unsigned int foo(struct st *st)
{
    return readl(&st->ptr);
}

mipsel-linux-gcc-4.5.1 -O2 foo.c -S output:
    lwl    $2,3($4)
    nop
    lwr    $2,0($4)
    j    $31
    nop
arm-linux-gnueabi-gcc-4.5.1 -O2 foo.c -S output:
    ldrb    r3, [r0, #0]    @ zero_extendqisi2
    ldrb    r1, [r0, #1]    @ zero_extendqisi2
    ldrb    r2, [r0, #2]    @ zero_extendqisi2
    orr    r3, r3, r1, asl #8
    ldrb    r0, [r0, #3]    @ zero_extendqisi2
    orr    r3, r3, r2, asl #16
    orr    r0, r3, r0, asl #24
    bx    lr

It seems the cast is ignored and unaligned access is assumed.
gcc 4.4.4 works fine.
mipsel-linux-gcc-4.4.4 -O2 foo.c -S output:
    lw    $2,0($4)
    j    $31
    nop
arm-linux-gnueabi-gcc-4.4.4 -O2 foo.c -S output:
    ldr    r0, [r0, #0]
    bx    lr

The similar problem was found as PR 45704.

git-bisect tell me same commit
0d9f1189f3df5ce5c0efc3ecadc7c0a4f75b202d is the first bad commit.

Like PR 45704, reverting this change fixes this problme too.

-  STRIP_USELESS_TYPE_CONVERSION (sub);
+  STRIP_NOPS (sub);

But the final fix for PR 45704 does not fix this problem.

mipsel-linux-gcc (4.6.0 20100927):
    lbu    $5,0($4)
    lbu    $6,1($4)
    lbu    $3,2($4)
    andi    $6,$6,0x00ff
    lbu    $2,3($4)
    andi    $5,$5,0x00ff
    sll    $6,$6,8
    andi    $3,$3,0x00ff
    or    $4,$6,$5
    sll    $3,$3,16
    or    $3,$3,$4
    sll    $2,$2,24
    j    $31
    or    $2,$2,$3
arm-linux-gnueabi-gcc (4.6.0 20100927):
    ldrb    r3, [r0, #0]    @ zero_extendqisi2
    ldrb    r1, [r0, #1]    @ zero_extendqisi2
    ldrb    r2, [r0, #2]    @ zero_extendqisi2
    orr    r3, r3, r1, asl #8
    ldrb    r0, [r0, #3]    @ zero_extendqisi2
    orr    r3, r3, r2, asl #16
    orr    r0, r3, r0, asl #24
    bx    lr


             reply	other threads:[~2010-09-28 15:30 UTC|newest]

Thread overview: 17+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-09-28 17:51 anemo at mba dot ocn.ne.jp [this message]
2010-09-28 18:38 ` [Bug c/45819] " rguenth at gcc dot gnu.org
2010-09-28 19:53 ` anemo at mba dot ocn.ne.jp
2010-09-30 10:00 ` [Bug middle-end/45819] " anemo at mba dot ocn.ne.jp
2010-09-30 11:56 ` rguenth at gcc dot gnu.org
2010-11-12 15:11 ` rguenth at gcc dot gnu.org
2010-11-24 14:43 ` anemo at mba dot ocn.ne.jp
2010-12-16 13:07 ` rguenth at gcc dot gnu.org
2011-02-22 13:40 ` anemo at mba dot ocn.ne.jp
2011-04-28 14:57 ` rguenth at gcc dot gnu.org
2011-06-09  6:47 ` raj.khem at gmail dot com
2011-07-22  9:50 ` rguenth at gcc dot gnu.org
2011-07-22 11:55 ` rguenth at gcc dot gnu.org
2011-07-22 12:20 ` rguenth at gcc dot gnu.org
2011-07-22 12:20 ` rguenth at gcc dot gnu.org
2011-07-22 12:22 ` rguenth at gcc dot gnu.org
2011-07-22 12:25 ` rguenth at gcc dot gnu.org

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=bug-45819-4@http.gcc.gnu.org/bugzilla/ \
    --to=gcc-bugzilla@gcc.gnu.org \
    --cc=gcc-bugs@gcc.gnu.org \
    /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).