Fix some BBRAM return values (from David Sidrane). Also some MTD-related cosmetic changes

This commit is contained in:
Gregory Nutt
2015-03-05 08:08:11 -06:00
parent 5158e10bc2
commit b6b50f5f42
+10 -4
View File
@@ -807,9 +807,9 @@ int stm32_bbsram_savepanic(int fileno, uint8_t *context, int length)
{
FAR struct bbsramfh_s *bbf;
int fill;
int ret = -ENOMEM;
int ret = -ENOSPC;
/* on a bad day we could panic while panicking, (and we debug assert)
/* On a bad day we could panic while panicking, (and we debug assert)
* this is a potential feeble attempt at only writing the first
* panic's context to the file
*/
@@ -826,13 +826,19 @@ int stm32_bbsram_savepanic(int fileno, uint8_t *context, int length)
DEBUGASSERT(bbf);
/* As once ensures we will keep the first dump checking the time for
/* If the g_bbsram has been nulled out we return ENXIO.
*
* As once ensures we will keep the first dump. Checking the time for
* 0 protects from over writing a previous crash dump that has not
* been saved to long term storage and erased. The dreaded reboot
* loop.
*/
if (bbf && (bbf->lastwrite.tv_sec == 0 && bbf->lastwrite.tv_nsec == 0))
if (!bbf)
{
ret = -ENXIO;
}
else if ((bbf->lastwrite.tv_sec == 0 && bbf->lastwrite.tv_nsec == 0))
{
/* Clamp length if too big */