Messages in this thread | | | Subject | Re: Alpha: virt_to_bus/GFP_DMA problem | Date | Sat, 11 Dec 1999 19:26:17 +0000 (GMT) | From | Alan Cox <> |
| |
> Even if that's the "workaround" for now, I would really prefer to have > a DMA_24 flag (replacing the current ISA DMA which can reach 24 bits > of addressing), and an identically defined 30, 31 and 32bit (and 28, > 26?), so that the code is clearly marked.
Its a very bad idea to start assuming nbit. It isnt that simple as chips may have memory windows. GFP_ISADMA maybe.
> Machines with ISA Olicom cards, often run out of DMA memory. We really > want to allocate a new buffer on every packet that comes in. That way
Sounds odd. You want to write tr_copy_and_checksum(). THe copy from the DMA ram to main memory (and more importantly L1 cache) has the same cost as the later checksum pass so you can do that instead of sucking up DMA memory.
> breaking down. Now, there have been improvements in this area, but > it sure was tricky to get right....
There are reasons that dev_alloc_skb() has no provision to force ISA memory. You end up with it failing, with sound cards unable to load and other bad stuff if you go allocating skbuffs from ISA pools
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |