[linux-dvb] Re: linux-2.6.0-test8 with dvb-kernel crashes

[Laurence Culhane <dvb@holmes.demon.co.uk>]

Hi,

I came across this problem a few months ago.  I do have more than 1GB of
RAM so felt I should be enabling the high memory support.  I've been
commenting out the BUG call that causes the problem (see below) with no
ill effects.
I'd be happier knowing that this was the correct thing to do but I have
to confess I haven't had the motivation to look into it further.  Sorry!

Laurence


/usr/src/linux-2.6.0-test7/drivers/media/common/saa7146_core.c
---
static struct scatterlist* vmalloc_to_sg(unsigned char *virt, int
nr_pages)
{
        struct scatterlist *sglist;
        struct page *pg;
        int i;

        sglist = kmalloc(sizeof(struct scatterlist)*nr_pages,
GFP_KERNEL);
        if (NULL == sglist)
                return NULL;
        memset(sglist,0,sizeof(struct scatterlist)*nr_pages);
        for (i = 0; i < nr_pages; i++, virt += PAGE_SIZE) {
                pg = vmalloc_to_page(virt);
                if (NULL == pg)
                        goto err;
/********
  For the time being ignore the test to see if we got some highmem
                if (PageHighMem(pg))
                        BUG();
********/
                sglist[i].page   = pg;
                sglist[i].length = PAGE_SIZE;
        }
        return sglist;
 



On Thu, 2003-10-23 at 17:25, Michael Hunold wrote:

> > This begs the question, why is the saa7146 code incompatible with high
> > memory set to 4GB?
> 
> I haven't investigated this yet. Fact is that 32bit PCI cards can only 
> address 4Gb address space, perhaps enabling high memory support marks 
> lower pages as high memory, too, and then my code crashes. But this is 
> just a wild guess.
> 




-- 
Info:
To unsubscribe send a mail to ecartis@linuxtv.org with "unsubscribe linux-dvb" as subject.