Hello,
(Please Cc: me on reply - I am not on the list)
I decided to give the voluntary preempt patch a try, but I ran into some
problems. I share them with you, hoping to contribute in some way to the
process.
I decided to build my kernel the Debian way (make-kpkg), with the config
file from the 2.6.8 package as a starting point. In order to get it to
build, I had to disable:
CONFIG_SCSI_ADVANSYS
CONFIG_DVB_BT8XX
CONFIG_DRM_GAMMA
and set
CONFIG_4KSTACKS=y
When booting the new kernel, problems occur with the ethernet device
(e100), and, not surprisingly, networking fails. Any idea? The IRQ 177
looks rather strange to me...
Maarten
Oct 6 14:22:59 mtg62 kernel: requesting new irq thread for IRQ177...
Oct 6 14:22:59 mtg62 kernel: Badness in enable_irq at kernel/irq/manage.c:111
Oct 6 14:22:59 mtg62 kernel: [enable_irq+259/272] enable_irq+0x103/0x110
Oct 6 14:22:59 mtg62 kernel: [pg0+543215970/1069945856] e100_up+0x162/0x210 [e100]
Oct 6 14:22:59 mtg62 kernel: [pg0+543212944/1069945856] e100_intr+0x0/0x140 [e100]
Oct 6 14:22:59 mtg62 kernel: [pg0+543220636/1069945856] e100_open+0x2c/0x80 [e100]
Oct 6 14:22:59 mtg62 kernel: [dev_open+133/160] dev_open+0x85/0xa0
Oct 6 14:22:59 mtg62 kernel: [dev_mc_upload+36/80] dev_mc_upload+0x24/0x50
Oct 6 14:22:59 mtg62 kernel: [dev_change_flags+83/304] dev_change_flags+0x53/0x130
Oct 6 14:22:59 mtg62 kernel: [dev_load+37/112] dev_load+0x25/0x70
Oct 6 14:22:59 mtg62 kernel: [devinet_ioctl+635/1712] devinet_ioctl+0x27b/0x6b0
Oct 6 14:22:59 mtg62 kernel: [inet_ioctl+102/176] inet_ioctl+0x66/0xb0
Oct 6 14:22:59 mtg62 kernel: [sock_ioctl+217/656] sock_ioctl+0xd9/0x290
Oct 6 14:22:59 mtg62 kernel: [sys_ioctl+234/592] sys_ioctl+0xea/0x250
Oct 6 14:22:59 mtg62 kernel: [syscall_call+7/11] syscall_call+0x7/0xb
Oct 6 14:22:59 mtg62 kernel: e100: eth0: e100_watchdog: link up, 100Mbps, full-duplex
Oct 6 14:22:59 mtg62 kernel: IRQ#177 thread started up.
On Wed, 2004-10-06 at 07:55, Maarten de Boer wrote:
> When booting the new kernel, problems occur with the ethernet device
> (e100), and, not surprisingly, networking fails. Any idea?
This is a known problem.
The maintainers are working on a solution.
It is only a warning and the e100 will
continue to work.
--
Paul Fulghum
[email protected]
> This is a known problem.
> The maintainers are working on a solution.
Ok.
> It is only a warning and the e100 will
> continue to work.
Hm, for me it didn't. To be more precise, configuring networking with
DHCP failed. But maybe that problem is somewhere else.
maarten
On Wed, 2004-10-06 at 08:10, Maarten de Boer wrote:
> > It is only a warning and the e100 will
> > continue to work.
>
> Hm, for me it didn't. To be more precise, configuring networking with
> DHCP failed. But maybe that problem is somewhere else.
The warning is caused by calling enable_irq()
when the interrupt enable depth is already 0
(interrupt is already enabled).
A couple of people, including me, have reported
this warning. The adapter continues to work.
My understanding of the problem is that it
is a warning only.
Do you see the interrupt count from /proc/interrupts
for the e100 increasing?
--
Paul Fulghum
[email protected]