Does this help at all: at any rate, it is a bit of the history of the
u-boot code:
http://lists.denx.de/pipermail/u-boot/2008-August/039679.html
Stephen Irons
David Brownell wrote:
I was looking at the DaVinci NAND support in current U-Boot
code (i.e. 2009.03 plus patches merged since that
Nori, Sekhar wrote:
From: Stephen Irons [mailto:stephen.ir...@tait.co.nz]
Sent: Thursday, April 16, 2009 10:37 AM
This is my preferred solution -- the appended patch is different from
the previous one.
1. I believe that emac_rx_bdproc() and emac_tx_bodproc() only set the
pending flag if
Nori, Sekhar wrote:
From: Stephen Irons
Sent: Tuesday, April 14, 2009 6:31 AM
Do not re-enable interrupts if receive processing has used its entire
budget.
As described in net/core/dev.c, function net_rx_action, drivers must
not
modify
the NAPI state if they consume their entire weight
review of
emac_rx_bdproc() and emac_poll() to those better qualified.
Stephen Irons
Stephen Irons wrote:
Do not re-enable interrupts if receive processing has used its entire
budget.
As described in net/core/dev.c, function net_rx_action, drivers must
not modify
the NAPI state if they con
Do not re-enable interrupts if receive processing has used its entire
budget.
As described in net/core/dev.c, function net_rx_action, drivers must not
modify
the NAPI state if they consume their entire weight. Davinci EMAC driver was
re-enabling interrupts in the rare case that it has used its
Nori, Sekhar wrote:
From: Stephen Irons
Sent: Tuesday, March 24, 2009 4:18 AM
As reported a few days ago, my DVEVM was producing a kernel panic with kernel
version 2.6.25-davinci1, 2.6.26-rc5 and 2.6.27-davinci1. It was consistently in
net_rx_action. It would happen after anything from a few
kely(budget <= 0 || jiffies != start_time))
goto softnet_break;
I hope these patches are usable: they are based on 2.6.27-davinci1 (+
some patches), not the latest head, and were generated using svn diff...
Stephen Irons
Senior Engineer
Tait Electronics Ltd
Index: drivers
kernel 2.6.28 or 2.6.29. In the
descriptions below, line numbers refer to the head of Davinci GIT.
This email is a description of the problem and some solutions. Patches
will come shortly.
Stephen Irons
Senior Engineer
Tait Electronics Ltd
Function emac_poll(), around line 2189:
Original code
, but can occur at any
time; once it happened during boot, before the kernel had finished loading.
It does not seem to happen when the network is connected via an old,
10Mbps hub.
Have you any insight to offer?
Stephen Irons
Senior Engineer
Tait Electronics Ltd
Christchurch
New Zealand
r