This is the mail archive of the ecos-discuss@sources.redhat.com mailing list for the eCos project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: stress_threads: I'm getting "Bad byte in chunk" output.Do I have memory corruption?


Andrew Lunn wrote:
We had similar problems on the EBSA285. It turned out to be a bug in
initializing the SDRAM controller. Both the HAL and the kernel have
tests programs for the cache. Try running those.

Hugo: If i remember correctly, you wrote a test program for this
bug. Is it one of these cache tests or is it another program?
I think it was hal/arm/ebsa285/current/tests/sdram0.cxx

On Tue, Nov 05, 2002 at 11:45:06AM -0500, Warren Postma wrote:

I'm getting an interesting bit of output when I run stress_threads.exe

This is the test code in stress_threads.c (circa line 600)
where perform_stressful_tasks() is cleaning up:

...
 /* now free it all up */
 for (i = 0; i < MAX_MALLOCED_SPACES; ++i) {
     if (spaces[i] != NULL) {
         size = sizes[i];
         for (j = 0; j < size; ++j) {
             // Validate chunk data.
             if ((0x50 | ((j+i) & 0x0f)) != spaces[i][j]) {
                 printf("Bad byte in chunk\n");
             }
             spaces[i][j] = 0xAA;    /* write a bit pattern */
         }
         free(spaces[i]);
     } else {
         ++statistics.malloc_failures;
     }
 }
...

In this case, I'm getting that "Bad byte in chunk" output to screen when I
run stress_threads? Is this likely to be a bug in my ECOS kernel or toolset?
Has anyone else seen this while running this test? I'm going to switch to a
different bunch of tests and see what other problems I can find on my
target.  I picked stress_threads as I thought, if the more brutal tests
pass, then its likely ECOS is stable on my target. <grin>
It could be that the heap memory that malloc thinks it can use (from your memory layout spec) doesn't match reality. You can also try the heaptest test from the memalloc package.

Jifl
--
eCosCentric http://www.eCosCentric.com/ <info@eCosCentric.com>
--[ "You can complain because roses have thorns, or you ]--
--[ can rejoice because thorns have roses." -Lincoln ]-- Opinions==mine


--
Before posting, please read the FAQ: http://sources.redhat.com/fom/ecos
and search the list archive: http://sources.redhat.com/ml/ecos-discuss


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]