sometimes get "crc error" while uncompressed ramdisk

sometimes get "crc error" while uncompressed ramdisk

[zhuzhenhua]

i make a ramdisk by myself, and sometimes the kernel boot the ramdisk
correctly but sometimes it printk "crc error" while uncompressed
ramdisk, did someone meet this situation?
thanks for any hints

Best regards


zhuzhenhua

Re: sometimes get "crc error" while uncompressed ramdisk

[jp]

On Fri, 2006-01-06 at 14:40 +0800, zhuzhenhua wrote:
> i make a ramdisk by myself, and sometimes the kernel boot the ramdisk
> correctly but sometimes it printk "crc error" while uncompressed
> ramdisk, did someone meet this situation?
> thanks for any hints
> 

Assuming your build and everything else is as it should be it may be a
RAM fault. Are you using a custom board?

I had some prototype boards here with some really long tracks to RAM.
(and some really short ones too!)

Memory tests such walking ones worked fine but the decompress of the
ramdisk works RAM pretty hard and it showed up intermittent faults like
you describe. Tended to be worse when the board was warm. I'd spray some
Freezit on and it would go back to working OK.

You could also try running the RAM slower and see if the fault
disappears.


JP

Re: sometimes get "crc error" while uncompressed ramdisk

[zhuzhenhua]

On 1/6/06, jp <jaypee@hotpop.com> wrote:
> On Fri, 2006-01-06 at 14:40 +0800, zhuzhenhua wrote:
> > i make a ramdisk by myself, and sometimes the kernel boot the ramdisk
> > correctly but sometimes it printk "crc error" while uncompressed
> > ramdisk, did someone meet this situation?
> > thanks for any hints
> >
>
> Assuming your build and everything else is as it should be it may be a
> RAM fault. Are you using a custom board?
>
> I had some prototype boards here with some really long tracks to RAM.
> (and some really short ones too!)
>
> Memory tests such walking ones worked fine but the decompress of the
> ramdisk works RAM pretty hard and it showed up intermittent faults like
> you describe. Tended to be worse when the board was warm. I'd spray some
> Freezit on and it would go back to working OK.
>
> You could also try running the RAM slower and see if the fault
> disappears.
>
>
> JP
>
>
>
maybe you are right, because i work on our FPGA board, and the
situation is similiar as your description. i will try to slow down the
sdram and to see what will happen

thanks!

Best regards

zhuzhenhua

Re: sometimes get "crc error" while uncompressed ramdisk

[zhuzhenhua]

and i find the ramdisk sometimes uncompressed correctly, sometimes not.
if if uncorrect, if often stop at
generic_file_write_nolock -> __grab_cache_page
->add_to_page_cache_unique
->__add_to_page_cache->add_page_to_hash_queue

static void add_page_to_hash_queue(struct page * page, struct page **p)
{
	struct page *next = *p;

	*p = page;
	page->next_hash = next;
	page->pprev_hash = p;
	if (next)
		next->pprev_hash = &page->next_hash;
	if (page->buffers)
		PAGE_BUG(page);
~~~~~~~~~~~~~~~~~~~~~~~~~~
	atomic_inc(&page_cache_size);
}

it occur oops at
	if (page->buffers)
		PAGE_BUG(page);

i am not sure it may caused by hardware or uncorrect initialized of sdram?
or maybe uncorrect-maked ramdisk can cause this?

thanks for any hints!

Best regards!

Zhuzhenhua




On 1/7/06, zhuzhenhua <zzh.hust@gmail.com> wrote:
> On 1/6/06, jp <jaypee@hotpop.com> wrote:
> > On Fri, 2006-01-06 at 14:40 +0800, zhuzhenhua wrote:
> > > i make a ramdisk by myself, and sometimes the kernel boot the ramdisk
> > > correctly but sometimes it printk "crc error" while uncompressed
> > > ramdisk, did someone meet this situation?
> > > thanks for any hints
> > >
> >
> > Assuming your build and everything else is as it should be it may be a
> > RAM fault. Are you using a custom board?
> >
> > I had some prototype boards here with some really long tracks to RAM.
> > (and some really short ones too!)
> >
> > Memory tests such walking ones worked fine but the decompress of the
> > ramdisk works RAM pretty hard and it showed up intermittent faults like
> > you describe. Tended to be worse when the board was warm. I'd spray some
> > Freezit on and it would go back to working OK.
> >
> > You could also try running the RAM slower and see if the fault
> > disappears.
> >
> >
> > JP
> >
> >
> >
> maybe you are right, because i work on our FPGA board, and the
> situation is similiar as your description. i will try to slow down the
> sdram and to see what will happen
>
> thanks!
>
> Best regards
>
> zhuzhenhua
>