Does anyone know how to do this?
All I need to do is, for a read request I need to read some sector off of
another block device. I'm using 2.4.18. I've been through the kernel
source and looked at the multi-device stuff. There are too many other
details to sort it all out. I just want to find some
help on creating buffer_head's, calling generic_make_request() and
completing the original buffer_head request when the request I created has
completed?
Does anyone know how to do this? If not, how do I found out? Do I need
to post to the kernel development mailing list?
This is what I've tried so far:
In my make_request() function I do this:
newBh = get_unused_buffer_head(1);
if (!newBh)
{
bh->b_end_io(bh, 0);
return 0;
}
memcpy(newBh, bh, sizeof(struct buffer_head));
newBh->b_rsector = sectorNum;
newBh->b_rdev = MKDEV(3, 0);
newBh->b_blocknr = newBh->b_rsector / 2; // My blocks are 512
bytes, hard disks are 1024
newBh->b_dev = newBh->b_rdev;
newBh->b_end_io = vdisk_end_io;
generic_make_request(rw, newBh);
return 0;
In my vdisk_end_io() function I do this:
struct buffer_head* masterBh = (struct buffer_head*)bh->b_private;
mark_buffer_uptodate(bh, uptodate);
masterBh->b_end_io(bh, uptodate);
put_unused_buffer_head(bh);
Thanks for any help.
On Fri, 23 Aug 2002, Jonathan Ludwig wrote:
> I want to call generic_make_request to do i/o on block drivers. I'm
> trying to figure out how to create buffer_head's to do i/o. I foung some
> code in the jbd source that creates a buffer_head for synchronous i/o; I
> want to do it asynchronously. I
> want to see if there is some documentation on how to do this. Does any
> one know of any documentation on this?
>
> I think it would go something like this:
>
> DoRead(unsigned long sector, unsinged numSectors, void* buffer)
> {
> struct page* newPage = virt_to_page(buffer);
> unsigned int newOffset = virt_to_offset(buffer);
> struct buffer_head* bh = get_unused_buffer_head(1);
> bh->b_prev_free = 0;
> bh->b_next_free = 0;
> init_buffer(bh, driver_end_io, NULL /* ?? Context ?? */);
> atomic_set(&bh->b_count, 1);
> set_bh_page(bh, newPage, newOffset);
> bh->b_size = numSectors; /* ?? bytes or sectors ?? */
> bh->b_dev = MAKEDEV(major, minor);
> bh->b_blocknr = sector;
> bh->b_state |= (1 << BH_Mapped) | (1 << BH_Dirty);
>
> generic_make_request(READ, bh);
> }
>
> I'm just guessing on this. Can anyone help me figure this out or refer me
> to a good source?
>
> Thanks
>
> --
> Kernelnewbies: Help each other learn about the Linux kernel.
> Archive: http://mail.nl.linux.org/kernelnewbies/
> FAQ: http://kernelnewbies.org/faq/
>
>
--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive: http://mail.nl.linux.org/kernelnewbies/
FAQ: http://kernelnewbies.org/faq/
