Re: [E-devel] Alignment trap on armv5tel

[Pierre C. <ca...@gm...>]

On 12/17/2010 11:12 AM, Cedric BAIL wrote:
> On Fri, Dec 17, 2010 at 11:04 AM, Pierre Cassimans <ca...@gm...> wrote:
>> On 12/17/2010 10:40 AM, Carsten Haitzler (The Rasterman) wrote:
>>> ooh.. wait.. you could be catching it at any point. no - you want to DISABLE
>>> alignment fixups in your kernel to make sure elementary_test segv's on
>>> unaligned access and have it running under gdb at the time.
>>
>> ok
>>
>> echo 5 > /proc/cpu/alignment did the trick.
>>
>> here is the new bt
>> Program received signal SIGBUS, Bus error.
>> _edje_part_recalc_single_aspect (ed=0x0, ep=0x0, desc=0x0,
>>    chosen_desc=0x906ec, rel1_to_x=0x150, rel1_to_y=0x0,
>> rel2_to_x=0x4048df10,
>>    rel2_to_y=0x2f6d6c65, confine_to=0x0, params=0x2f657361) at
>> edje_calc.c:427
>> 427     edje_calc.c: No such file or directory.
>>        in edje_calc.c
>> (gdb) bt
>> #0  _edje_part_recalc_single_aspect (ed=0x0, ep=0x0, desc=0x0,
>>    chosen_desc=0x906ec, rel1_to_x=0x150, rel1_to_y=0x0,
>> rel2_to_x=0x4048df10,
>>    rel2_to_y=0x2f6d6c65, confine_to=0x0, params=0x2f657361) at
>> edje_calc.c:427
>> #1  _edje_part_recalc_single (ed=0x0, ep=0x0, desc=0x0,
>> chosen_desc=0x906ec,
>>    rel1_to_x=0x150, rel1_to_y=0x0, rel2_to_x=0x4048df10,
>>    rel2_to_y=0x2f6d6c65, confine_to=0x0, params=0x2f657361)
>>    at edje_calc.c:1312
>> #2  0x4042df24 in _edje_part_recalc (ed=0x40022ee0, ep=0x912c8,
>>    flags=-1093060952) at edje_calc.c:1810
>> #3  0x40430e70 in _edje_recalc_do (ed=0x86ac0) at edje_calc.c:268
>> #4  0x4046928c in _edje_smart_resize (obj=<value optimized out>,
>>    w=<value optimized out>, h=1) at edje_smart.c:224
>> #5  0x4082a61c in evas_object_resize (obj=0x86998, w=1, h=1)
>>    at evas_object_main.c:511
>> #6  0x40050624 in _smart_reconfigure (sd=0x86858) at elm_widget.c:2425
>> #7  0x4082a61c in evas_object_resize (obj=0x86708, w=1, h=1)
>>    at evas_object_main.c:511
>> #8  0x4004b118 in elm_win_resize_object_add (obj=0x85bd8, subobj=0x86708)
>>    at elm_win.c:1127
>> #9  0x00013c44 in my_win_main (argc=<value optimized out>,
>>    argv=<value optimized out>) at test.c:201
>> #10 elm_main (argc=<value optimized out>, argv=<value optimized out>)
>>    at test.c:404
>> ---Type <return> to continue, or q <return> to quit---
>> #11 0x40d9b038 in __libc_start_main () from /lib/libc.so.6
>> #12 0x00013b00 in _start ()
> 
> That's really weird, I do have a target that highly dislike memory
> misalignment and do use the fixed point math, and I don't see that
> issue at all. Maybe you have some special configuration with eina
> memory pool. What is the content of eina configure report (the last
> few lines) ?

## ----------- ##
## confdefs.h. ##
## ----------- ##

/* confdefs.h */
#define PACKAGE_NAME "eina"
#define PACKAGE_TARNAME "eina"
#define PACKAGE_VERSION "1.0.0.beta2"
#define PACKAGE_STRING "eina 1.0.0.beta2"
#define PACKAGE_BUGREPORT "enl...@li..."
#define PACKAGE_URL ""
#define STDC_HEADERS 1
#define HAVE_SYS_TYPES_H 1
#define HAVE_SYS_STAT_H 1
#define HAVE_STDLIB_H 1
#define HAVE_STRING_H 1
#define HAVE_MEMORY_H 1
#define HAVE_STRINGS_H 1
#define HAVE_INTTYPES_H 1
#define HAVE_STDINT_H 1
#define HAVE_UNISTD_H 1
#define __EXTENSIONS__ 1
#define _ALL_SOURCE 1
#define _GNU_SOURCE 1
#define _POSIX_PTHREAD_SEMANTICS 1
#define _TANDEM_SOURCE 1
#define PACKAGE "eina"
#define VERSION "1.0.0.beta2"
#define HAVE_DLFCN_H 1
#define LT_OBJDIR ".libs/"
#define VMAJ 1
#define VMIN 0
#define VMIC 0
#define VREV 0
#define MODULE_ARCH "linux-gnueabi-arm-1.0.0"
#define SHARED_LIB_SUFFIX ".so"
#define EFL_HAVE_POSIX_THREADS 1
#define EFL_HAVE_THREADS 1
#define EFL_HAVE_POSIX_THREADS_SPINLOCK 1
#define NVALGRIND 1
#define EINA_SAFETY_CHECKS 1
#define HAVE_DIRENT_H 1
#define TIME_WITH_SYS_TIME 1
#define SIZEOF_WCHAR_T 4
#define HAVE___ATTRIBUTE__ 1
#define __UNUSED__ __attribute__((unused))
#define HAVE_ALLOCA_H 1
#define HAVE_ALLOCA 1
#define HAVE_DLADDR 1
#define HAVE_ICONV_H 1
#define HAVE_ICONV 1
#define HAVE_DIRFD 1
#define HAVE_OPENAT 1
#define HAVE_OPENAT 1
#define HAVE_FPATHCONF 1
#define HAVE_FPATHCONF 1
#define EINA_STATIC_BUILD_CHAINED_POOL 1
#define EINA_BUILD_CHAINED_POOL 1
#define EINA_STATIC_BUILD_ONE_BIG 1
#define EINA_BUILD_ONE_BIG 1

and for the libc version:

how can i check which versions of the toolchain are used?

also, i see in the config file for the distro that
LINUX_LIBC_HEADERS_VERSION ?= "2.6.31" and my kernel is 2.6.21-hh20 <
can this be a problem?

and glibc is version 2.9