From: Francois Romieu <[EMAIL PROTECTED]>
Date: Wed, 7 Sep 2005 20:59:49 +0200
> Stephen Hemminger <[EMAIL PROTECTED]> :
> [...]
> > It really sounds like netconsole needs to have a different device hook
> > instead of start_xmit. It also probably doesn't want to have allocate
> > an skb. What y
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Wed, 7 Sep 2005 10:04:29 -0700
> Netpoll is intended to be a replacement for the requirement of serial
> cables for doing things like tricky oops capture and kgdb.
>
> If we move to softirq-only processing, it will not be a serial cable
> replacement.
On Wed, Sep 07, 2005 at 12:17:01PM -0700, Ben Greear wrote:
> Francois Romieu wrote:
> >Stephen Hemminger <[EMAIL PROTECTED]> :
> >[...]
> >
> >>It really sounds like netconsole needs to have a different device hook
> >>instead of start_xmit. It also probably doesn't want to have allocate
> >>an s
Francois Romieu wrote:
Stephen Hemminger <[EMAIL PROTECTED]> :
[...]
It really sounds like netconsole needs to have a different device hook
instead of start_xmit. It also probably doesn't want to have allocate
an skb. What you want is a synchronous way to send one packet with
interrupts disab
Stephen Hemminger <[EMAIL PROTECTED]> :
[...]
> It really sounds like netconsole needs to have a different device hook
> instead of start_xmit. It also probably doesn't want to have allocate
> an skb. What you want is a synchronous way to send one packet with
> interrupts disabled:
>
> (de
On Wed, Sep 07, 2005 at 11:19:42AM -0700, Stephen Hemminger wrote:
> On Wed, 7 Sep 2005 10:19:33 -0700
> Matt Mackall <[EMAIL PROTECTED]> wrote:
>
> > On Tue, Sep 06, 2005 at 09:02:42PM -0700, Eugene Surovegin wrote:
> > > I don't quite understand, you _already_ have deferred processing in
> > >
On Wed, 7 Sep 2005 10:19:33 -0700
Matt Mackall <[EMAIL PROTECTED]> wrote:
> On Tue, Sep 06, 2005 at 09:02:42PM -0700, Eugene Surovegin wrote:
> > I don't quite understand, you _already_ have deferred processing in
> > netpoll (btw, how good this will work with kgdboe?).
>
> Yes, it exists as a l
Matt Mackall wrote:
> On Wed, Sep 07, 2005 at 07:16:03PM +0200, Patrick McHardy wrote:
>
>>Not sure if its possible, but perhaps you could register the panic
>>notifier to do whatever is necessary to get an oops out.
>
>
> It's not really just about oopses.
>
> Consider the case of:
>
>
On Wed, Sep 07, 2005 at 07:16:03PM +0200, Patrick McHardy wrote:
> Matt Mackall wrote:
> > If we move to softirq-only processing, it will not be a serial cable
> > replacement. It won't capture the tricky oopses you normally need a
> > serial cable for and it won't work with kgdb. It will be an awk
On Tue, Sep 06, 2005 at 09:02:42PM -0700, Eugene Surovegin wrote:
> I don't quite understand, you _already_ have deferred processing in
> netpoll (btw, how good this will work with kgdboe?).
Yes, it exists as a last ditch fallback for cases where we can detect
deadlock. And it works not at all fo
Matt Mackall wrote:
> If we move to softirq-only processing, it will not be a serial cable
> replacement. It won't capture the tricky oopses you normally need a
> serial cable for and it won't work with kgdb. It will be an awkward
> syslogd and nothing more. I won't have any use for it and I suspec
On Tue, Sep 06, 2005 at 08:42:39PM -0700, David S. Miller wrote:
> You'll need to find a way to do netpoll without disabling hw IRQs
> during these driver methods. Then netpoll would be universally
> usable, as it would even work with drivers which are software tunnels
> and which feed packets bac
On Tue, Sep 06, 2005 at 08:49:35PM -0700, Matt Mackall wrote:
> On Tue, Sep 06, 2005 at 08:37:36PM -0700, Eugene Surovegin wrote:
> > On Tue, Sep 06, 2005 at 08:22:25PM -0700, Matt Mackall wrote:
> > > Which would I rather have:
> > >
> > > "netconsole never catches my oopses, it's useless."
> >
On Tue, Sep 06, 2005 at 08:37:36PM -0700, Eugene Surovegin wrote:
> On Tue, Sep 06, 2005 at 08:22:25PM -0700, Matt Mackall wrote:
> > Which would I rather have:
> >
> > "netconsole never catches my oopses, it's useless."
> >
> > "netconsole didn't work with my driver, so I tried another card and
On Tue, Sep 06, 2005 at 08:35:32PM -0700, Matt Mackall wrote:
> On Tue, Sep 06, 2005 at 08:19:34PM -0700, David S. Miller wrote:
> > From: Matt Mackall <[EMAIL PROTECTED]>
> > Date: Tue, 6 Sep 2005 20:08:10 -0700
> >
> > > Think upon the kgdb-over-ethernet case, please. The kernel hits a
> > > bre
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 20:35:32 -0700
> Kgdboe simply wants the asynchronous subsystem to get out of the
> way. For most cards, it works just fine.
It wants to impose a locking model restriction which never ever
existed in the past.
As I said, I myself even
On Tue, Sep 06, 2005 at 08:22:25PM -0700, Matt Mackall wrote:
> Which would I rather have:
>
> "netconsole never catches my oopses, it's useless."
>
> "netconsole didn't work with my driver, so I tried another card and it
> works great."
Well, not all world which uses Linux is PC with PCI slots
On Tue, Sep 06, 2005 at 08:19:34PM -0700, David S. Miller wrote:
> From: Matt Mackall <[EMAIL PROTECTED]>
> Date: Tue, 6 Sep 2005 20:08:10 -0700
>
> > Think upon the kgdb-over-ethernet case, please. The kernel hits a
> > breakpoint, the kgdb stub stops everything, sends a packet to the
> > debuggi
On Tue, Sep 06, 2005 at 07:32:38PM -0700, Eugene Surovegin wrote:
> On Tue, Sep 06, 2005 at 07:19:21PM -0700, Matt Mackall wrote:
> > > or a') make this a per-driver feature (e.g. NETIF_F_NETPOLL_CHALENGED)
> > >
> > > In this case, even if driver cannot handle being called from IRQ
> > > context
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 20:08:10 -0700
> Think upon the kgdb-over-ethernet case, please. The kernel hits a
> breakpoint, the kgdb stub stops everything, sends a packet to the
> debugging client, waits for a packet back.. This simply can't work if
> we delay sen
On Tue, Sep 06, 2005 at 07:37:45PM -0700, David S. Miller wrote:
> From: Matt Mackall <[EMAIL PROTECTED]>
> Date: Tue, 6 Sep 2005 18:03:38 -0700
>
> > Option c) is obviously a big project but maybe we can get from here to
> > there. One possible step in that direction would be exposing a
> > stand
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 18:03:38 -0700
> Option c) is obviously a big project but maybe we can get from here to
> there. One possible step in that direction would be exposing a
> standard driver lock that netpoll can see and switching drivers that
> have troubl
On Tue, Sep 06, 2005 at 07:19:21PM -0700, Matt Mackall wrote:
> > or a') make this a per-driver feature (e.g. NETIF_F_NETPOLL_CHALENGED)
> >
> > In this case, even if driver cannot handle being called from IRQ
> > context, netconsole still can be used, although in a little more
> > limited fashi
On Tue, Sep 06, 2005 at 06:37:40PM -0700, Eugene Surovegin wrote:
> On Tue, Sep 06, 2005 at 06:03:38PM -0700, Matt Mackall wrote:
> > On Tue, Sep 06, 2005 at 04:01:24PM -0700, David S. Miller wrote:
> > > So you cannot call into these drivers with HW interrupts disabled or
> > > even worse from HW
On Tue, Sep 06, 2005 at 06:03:38PM -0700, Matt Mackall wrote:
> On Tue, Sep 06, 2005 at 04:01:24PM -0700, David S. Miller wrote:
> > So you cannot call into these drivers with HW interrupts disabled or
> > even worse from HW interrupt context. These drivers use locking
> > strategies which are per
On Tue, Sep 06, 2005 at 04:01:24PM -0700, David S. Miller wrote:
> So you cannot call into these drivers with HW interrupts disabled or
> even worse from HW interrupt context. These drivers use locking
> strategies which are perfectly legal and work until you add netpoll.
And again, I agree.
Wha
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 15:50:25 -0700
> > So we try to spit out netconsole messages in hw IRQ context and stuff
> > like that, as you stated. The tg3 driver is susceptible to the
> > problem you mention, as is bnx2, because they use purely software
> > interr
On Tue, Sep 06, 2005 at 03:42:07PM -0700, David S. Miller wrote:
> Mentioning the latency of the serial console, in support of
> netpoll's interrupt disabling, is quite a straw man.
No, it's exactly to the point: latency is a secondary concern when
we're printing an oops or other diagnostic. Other
On Tue, Sep 06, 2005 at 03:36:27PM -0700, David S. Miller wrote:
> From: Eugene Surovegin <[EMAIL PROTECTED]>
> Date: Tue, 6 Sep 2005 15:04:17 -0700
>
> > David, correct me if I'm wrong, but I think there is a major problem
> > with current netconsole/netpoll approach.
>
> You're preaching to the
From: Matt Mackall <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 15:32:50 -0700
> Unfortunately, netpoll fundamentally requires everything to work with
> interrupts off. This can't be changed. It could be called from the
> context of an oops, or worse, by the kgdb stub at a breakpoint. If
> drivers re
On Tue, Sep 06, 2005 at 03:36:27PM -0700, David S. Miller wrote:
> From: Eugene Surovegin <[EMAIL PROTECTED]>
> Date: Tue, 6 Sep 2005 15:04:17 -0700
>
> > David, correct me if I'm wrong, but I think there is a major problem
> > with current netconsole/netpoll approach.
>
> You're preaching to the
From: Eugene Surovegin <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 15:04:17 -0700
> David, correct me if I'm wrong, but I think there is a major problem
> with current netconsole/netpoll approach.
You're preaching to the choir. I think the whole netpoll
implementation is fundamentally flawed, and
On Tue, Sep 06, 2005 at 02:47:14PM -0700, David S. Miller wrote:
> From: Eugene Surovegin <[EMAIL PROTECTED]>
> Date: Tue, 6 Sep 2005 00:40:01 -0700
>
> > According to Documentation/networking/netdevices.txt
> > dev->hard_start_xmit must be called with interrupts *enabled*.
> >
> > Unfortunately
On Tue, Sep 06, 2005 at 02:47:14PM -0700, David S. Miller wrote:
> From: Eugene Surovegin <[EMAIL PROTECTED]>
> Date: Tue, 6 Sep 2005 00:40:01 -0700
>
> > According to Documentation/networking/netdevices.txt
> > dev->hard_start_xmit must be called with interrupts *enabled*.
> >
> > Unfortunately
From: Eugene Surovegin <[EMAIL PROTECTED]>
Date: Tue, 6 Sep 2005 00:40:01 -0700
> According to Documentation/networking/netdevices.txt
> dev->hard_start_xmit must be called with interrupts *enabled*.
>
> Unfortunately, current netconsole code always calls netpoll with local
> interrupts disable
35 matches
Mail list logo
