[Bug c/67223] New: Address misaligned for ldrexd

[Bug c/67223] New: Address misaligned for ldrexd

[soldier.kam at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67223

            Bug ID: 67223
           Summary: Address misaligned for ldrexd
           Product: gcc
           Version: 5.2.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
          Assignee: unassigned at gcc dot gnu.org
          Reporter: soldier.kam at gmail dot com
  Target Milestone: ---

Hi!
I compiled wrk (https://github.com/wg/wrk) for odroid-c1. Binary runs but cause
bus error. I thought that it is caused by ldrexd that requires address to be
doubleword aligned but it is not - it is 0xb5c3103c. See below for details.

In dmesg I found: 
"Alignment trap: wrk (19927) PC=0x00017494 Instr=0xe1b14f9f Address=0xb5c3212c
FSR 0x001"

GDBs "show registers":
r0             0x1e8481 2000001
r1             0xb5c3103c       3049459772
r2             0x1      1
r3             0x0      0
r4             0x16378  91000
r5             0xbefffa08       3204446728
r6             0x0      0
r7             0x152    338
r8             0xbefffa08       3204446728
r9             0x0      0
r10            0xb6ff94c0       3070203072
r11            0xb0f85cfc       2969066748
r12            0xb5c3000c       3049455628
sp             0xb0f85cd0       0xb0f85cd0
lr             0x15fcc  90060
pc             0x17494  0x17494 <stats_record+84>
cpsr           0x200b0010       537591824

Assembler for this function:
00017440 <stats_record>:
   17440:       e92d4830        push    {r4, r5, fp, lr}
   17444:       e28db00c        add     fp, sp, #12
   17448:       e24dd020        sub     sp, sp, #32
   1744c:       e50b0020        str     r0, [fp, #-32]  ; 0xffffffe0
   17450:       e14b22fc        strd    r2, [fp, #-44]  ; 0xffffffd4
   17454:       e51b3020        ldr     r3, [fp, #-32]  ; 0xffffffe0
   17458:       e1c300d8        ldrd    r0, [r3, #8]
   1745c:       e14b22dc        ldrd    r2, [fp, #-44]  ; 0xffffffd4
   17460:       e1510003        cmp     r1, r3
   17464:       01500002        cmpeq   r0, r2
   17468:       8a000001        bhi     17474 <stats_record+0x34>
   1746c:       e3a03000        mov     r3, #0
   17470:       ea000048        b       17598 <stats_record+0x158>
   17474:       e51b302c        ldr     r3, [fp, #-44]  ; 0xffffffd4
   17478:       e2833004        add     r3, r3, #4
   1747c:       e1a03183        lsl     r3, r3, #3
   17480:       e51b2020        ldr     r2, [fp, #-32]  ; 0xffffffe0
   17484:       e0821003        add     r1, r2, r3
   17488:       e3a02001        mov     r2, #1
   1748c:       e3a03000        mov     r3, #0
   17490:       f57ff05f        dmb     sy
   17494:       e1b14f9f        ldrexd  r4, [r1]
   17498:       e0944002        adds    r4, r4, r2
   1749c:       e0a55003        adc     r5, r5, r3
   174a0:       e1a10f94        strexd  r0, r4, [r1]
   174a4:       e3500000        cmp     r0, #0
   174a8:       1afffff9        bne     17494 <stats_record+0x54>
   174ac:       f57ff05f        dmb     sy
   174b0:       e51b1020        ldr     r1, [fp, #-32]  ; 0xffffffe0
   174b4:       e3a02001        mov     r2, #1
   174b8:       e3a03000        mov     r3, #0
   174bc:       f57ff05f        dmb     sy
   174c0:       e1b14f9f        ldrexd  r4, [r1]
   174c4:       e0944002        adds    r4, r4, r2
   174c8:       e0a55003        adc     r5, r5, r3
   174cc:       e1a10f94        strexd  r0, r4, [r1]
   174d0:       e3500000        cmp     r0, #0
   174d4:       1afffff9        bne     174c0 <stats_record+0x80>
   ...

Output "gcc -v":
Using built-in specs.
COLLECT_GCC=gcc
COLLECT_LTO_WRAPPER=/usr/lib/gcc/armv7l-unknown-linux-gnueabihf/5.2.0/lto-wrapper
Target: armv7l-unknown-linux-gnueabihf
Configured with: /build/gcc/src/gcc-5.2.0/configure --prefix=/usr
--libdir=/usr/lib --libexecdir=/usr/lib --mandir=/usr/share/man
--infodir=/usr/share/info
--with-bugurl=https://github.com/archlinuxarm/PKGBUILDs/issues
--enable-languages=c,c++,fortran,go,lto,objc,obj-c++ --enable-shared
--enable-threads=posix --with-system-zlib --with-isl --enable-__cxa_atexit
--disable-libunwind-exceptions --enable-clocale=gnu --disable-libstdcxx-pch
--disable-libssp --enable-gnu-unique-object --enable-linker-build-id
--enable-lto --enable-plugin --enable-install-libiberty
--with-linker-hash-style=gnu --enable-gnu-indirect-function --disable-multilib
--disable-werror --enable-checking=release
--with-default-libstdcxx-abi=gcc4-compatible
--host=armv7l-unknown-linux-gnueabihf --build=armv7l-unknown-linux-gnueabihf
--with-arch=armv7-a --with-float=hard --with-fpu=vfpv3-d16
Thread model: posix
gcc version 5.2.0 (GCC)

[Bug target/67223] Address misaligned for ldrexd

[mikpelinux at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67223

--- Comment #1 from Mikael Pettersson <mikpelinux at gmail dot com> ---
You didn't provide a self-contained test case.  After extracting stats_record
and the stats declaration from the source to linked to and providing my own
stats_alloc (I'm not using your zcalloc), I cannot reproduce the SIGBUS. 
(Tried it on a Cortex-A9 system.)

The alignment of a stats struct is 8.  Verify that your allocator returns
correctly aligned objects.  And do provide a self-contained test case -- one
that compiles, runs, and provokes the crash.

[Bug target/67223] Address misaligned for ldrexd

[soldier.kam at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=67223

Marcin Kamionowski <soldier.kam at gmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |INVALID

--- Comment #2 from Marcin Kamionowski <soldier.kam at gmail dot com> ---
Yes, you are correct about misaligned address returned from zmalloc. Now it
works without errors. Thank you for helping out. I was not aware of zmalloc
features.