Re: tried ia64-net-20020506.iso, (rx4610) failed with map_single IO TLB kernel panic...
On Wed, May 22, 2002 at 05:36:07PM -0700, David Mosberger wrote:
> >>>>> On Wed, 22 May 2002 13:45:36 +0200, "Jiri Polach" <email@example.com> said:
> Jiri> To be more specific:
> Jiri> "Kernel panic: map_single: could not allocate software IO TLB
> Jiri> (1552 bytes)"
> Jiri> So strictly speaking I probably do not need swiotlb=32768, but
> Jiri> something like swiotlb=2048 should be sufficient. On the other
> Jiri> hand the default is too small.
> Yes, but I'm wondering why 1024 isn't enough. IIRC, the eepro100
> driver uses a tx/rx buffer that has at most 64 packets, so a software
> I/O TLB of 2MB should be plenty big. Also, I have never seen this
> problem myself, so I'm wondering whether this problem pops up only
> under certain circumstances.
I see this patch in the debian ia64 kernel sources, which sounds like it
might be relevant. No idea why it is there, but it is part of the hplso
diff -urN --exclude-from=/home/rh-ia64/ia64/redhat/dontdiff ../ia64/linux/drivers/net/eepro100.c linux/drivers/net/eepro100.c
--- ../ia64/linux/drivers/net/eepro100.c Fri Jan 25 13:11:22 2002
+++ linux/drivers/net/eepro100.c Wed Feb 27 15:19:22 2002
@@ -70,8 +70,8 @@
/* A few values that may be tweaked. */
/* The ring sizes should be a power of two for efficiency. */
-#define TX_RING_SIZE 32
-#define RX_RING_SIZE 32
+#define TX_RING_SIZE 64
+#define RX_RING_SIZE 1024
/* How much slots multicast filter setup may take.
Do not descrease without changing set_rx_mode() implementaion. */
#define TX_MULTICAST_SIZE 2
To UNSUBSCRIBE, email to firstname.lastname@example.org
with a subject of "unsubscribe". Trouble? Contact email@example.com