On Wed, Aug 03, 2005 at 21:07:46 -0400, Hemant Mohapatra wrote: > On 8/3/05, Vincenzo Mallozzi <vinjunior@xxxxxxxx> wrote: > I am not sure if this error could _also_ be caused by the wrong > kmalloc flag. Maybe someone more experienced in linux-mm could answer > that. I believe it would be different error. That is it would either complaint about trying to sleep in an interrupt, or simply lock up hard. The problem might be, that kmalloc in GFP_ATOMIC mode may quite easily return NULL. However an OOPS can much more probably be caused by actually accessing an invalid pointer. It can have the same causes as SIGSEGV does in userland. However in this specific case, hijacking a page fault handler, may cause it via other means -- it is *the page fault handler* that prints the message, so it can as well be some problem in chaining to the original hanler. Anyway, to get more precise answer, you must post the code. If it's too big, then put it on the web somewhere and post an URL. ------------------------------------------------------------------------------- Jan 'Bulb' Hudec <bulb@xxxxxx>
Attachment:
signature.asc
Description: Digital signature
