Libor> Yeah, I had the same memory come back when I saw the error
Libor> message. I'll go back and look to apply the patch Michael
Libor> submitted. Also does FMR registration currently perform the
Libor> dma mapping, or does the consumer need to do it?
The consumer needs to do it
On Thu, Jun 16, 2005 at 03:44:11PM -0700, Roland Dreier wrote:
> Troy> WARN: : Failed to find lock pointer. WARN: : Error <-22>
> Troy> initializing memory locking. WARN: : Error <-22>
> Troy> initializing SDP IOCB table. WARN: : INIT: Error <-22>
> Troy> initializing connection
We should be checking the return value of copy_from_user(), right?
--- infiniband/core/user_mad.c (revision 2641)
+++ infiniband/core/user_mad.c (working copy)
@@ -355,18 +355,26 @@ static ssize_t ib_umad_write(struct file
if (!rmpp_active) {
/* Copy message from user i
Troy> WARN: : Failed to find lock pointer. WARN: : Error <-22>
Troy> initializing memory locking. WARN: : Error <-22>
Troy> initializing SDP IOCB table. WARN: : INIT: Error <-22>
Troy> initializing connection table.
Oh yeah, the current SDP code has no chance of working on ppc64
> From: Grant Grundler [mailto:[EMAIL PROTECTED]
> Sent: Thursday, June 16, 2005 3:27 PM
>
> I don't agree that all the existing linux content should be
> mangled to include windows content on the same pages.
> Windows folks should clone any pages needing more than a few
> lines changed to accommo
> From: Roland Dreier [mailto:[EMAIL PROTECTED]
> Sent: Thursday, June 16, 2005 3:22 PM
>
> Fab> I can see how to edit pages, but don't see any links or
> Fab> buttons to create new pages... I'm likely just missing it.
> Fab> I'll look some more, but if you have a pointer, that would
Quoting r. Troy Benjegerdes <[EMAIL PROTECTED]>:
> Subject: Re: SDP: device mthca0 does not support fast memory regions
>
> On Thu, Jun 16, 2005 at 10:03:55PM +0300, Michael S. Tsirkin wrote:
> > Quoting r. Libor Michalek <[EMAIL PROTECTED]>:
> > > Subject: Re: SDP: device mthca0 does not support
On Thu, Jun 16, 2005 at 10:03:55PM +0300, Michael S. Tsirkin wrote:
> Quoting r. Libor Michalek <[EMAIL PROTECTED]>:
> > Subject: Re: SDP: device mthca0 does not support fast memory regions
> >
> > On Wed, Jun 15, 2005 at 04:30:47PM -0500, Troy Benjegerdes wrote:
> > > I'm getting a 'SDP: device m
On Thu, Jun 16, 2005 at 03:14:54PM -0700, Fab Tillier wrote:
> > I thought we wanted to keep the Windows project separate from the
> > Linux project and not share this short of thing though...
...
> I don't see why we have to keep the web content separate. In fact,
> since the Wiki is at the top l
Fab> I can see how to edit pages, but don't see any links or
Fab> buttons to create new pages... I'm likely just missing it.
Fab> I'll look some more, but if you have a pointer, that would be
Fab> great.
Actually I guess you have to create a link on an existing page first,
either
Michael> Why dont we just set VM_DONTCOPY?
Hmm, that's a good point. I think that solves the problem in a much
simpler way.
- R.
___
openib-general mailing list
openib-general@openib.org
http://openib.org/mailman/listinfo/openib-general
To unsubs
> From: Roland Dreier [mailto:[EMAIL PROTECTED]
> Sent: Thursday, June 16, 2005 2:55 PM
>
> Fab> This looks good. I'd like to setup a page for the windows
> Fab> project, rather than duplicating all of this infrastructure.
> Fab> How would I go about this?
>
> Just create an account
Fab> This looks good. I'd like to setup a page for the windows
Fab> project, rather than duplicating all of this infrastructure.
Fab> How would I go about this?
Just create an account on the wiki and then create whatever pages you
want (and edit existing pages to add links as necessar
Quoting r. Roland Dreier <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] Ref count user doorbell pages in mthca
>
> Michael> Looking at code under mm/, it seems that when you mmap
> Michael> something, get_file is called on the descriptor (and fput
> Michael> on close). So it seems to me
> From: Tom Duffy [mailto:[EMAIL PROTECTED]
> Sent: Thursday, June 16, 2005 2:04 PM
>
> On Tue, 2005-06-14 at 11:53 -0700, Matt Leininger wrote:
> > On Tue, 2005-06-14 at 11:38 -0700, Tom Duffy wrote:
> > > Can the folks at Sandia please update the http://www.openib.org/doc.html
> > > page to poin
Michael> Looking at code under mm/, it seems that when you mmap
Michael> something, get_file is called on the descriptor (and fput
Michael> on close). So it seems to me the driver's release wont
Michael> get called until the app actually unmaps the memory. If
Michael> thats ri
Quoting r. Roland Dreier <[EMAIL PROTECTED]>:
> Subject: Re: [PATCH] Ref count user doorbell pages in mthca
>
> Roland> This patch keeps a reference count on the doorbell pages that are
> Roland> mapped into userspace for direct verbs access. This prevents a
> Roland> buggy/malicious
On Tue, 2005-06-14 at 11:53 -0700, Matt Leininger wrote:
> On Tue, 2005-06-14 at 11:38 -0700, Tom Duffy wrote:
> > Can the folks at Sandia please update the http://www.openib.org/doc.html
> > page to point to the wiki site?
> >
> > Also, we should move the FAQ's to WIKI. I can do this if you want
Michael> How does one close the context? I thought thats done by
Michael> closing the file descriptor, isnt that right? And if not
Michael> - why not? I think closing the file descriptor will kill
Michael> the vmas without work in mthca, avoiding this issue
Michael> altogether.
Sean,
Here is a to fix a bug in service_id/service_mask
handling in CM.
Previously, if you tried to listen on service_id:
service_mask of 0x00:0x0F, followed by a listen
on 0xF0:0xF0, the second listen would fail because
the two ranges overlap (0xF0 appears in both ranges).
But, if you tried the
Quoting r. Roland Dreier <[EMAIL PROTECTED]>:
> Subject: [PATCH] Ref count user doorbell pages in mthca
>
> This patch keeps a reference count on the doorbell pages that are
> mapped into userspace for direct verbs access. This prevents a
> buggy/malicious app from closing its context but keeping
This patch keeps a reference count on the doorbell pages that are
mapped into userspace for direct verbs access. This prevents a
buggy/malicious app from closing its context but keeping a doorbell
page mapped, and then improperly accessing the doorbell page after it
gets allocated again to a new p
>Itamar,
>
>Can you seperate the RMR api change and hash table removal into a
>seperate patch?
I dont see why, the new input param to dat_rmr_bind (lmr_handle) come
in order that we will be able to remove the hash table so this is one
patch.
>Merging the HCA and IA code into the dapl_ia.[ch] fil
Quoting r. Libor Michalek <[EMAIL PROTECTED]>:
> Subject: Re: SDP: device mthca0 does not support fast memory regions
>
> On Wed, Jun 15, 2005 at 04:30:47PM -0500, Troy Benjegerdes wrote:
> > I'm getting a 'SDP: device mthca0 does nto support fast memory regions'
> > error on PPC64 systems.. Is th
Libor> I'm not sure about the state or plans for FMR support on
Libor> PPC64, Roland will have a better idea.
There's no reason it shouldn't work with the latest code. I've not
tried it, but the only reason that mthca would not provide the FMR
functions would be if the FW is burned with
On Wed, 2005-06-15 at 17:30, Troy Benjegerdes wrote:
> I'm getting a 'SDP: device mthca0 does nto support fast memory regions'
> error on PPC64 systems.. Is there something that needs to be done for
> PPC, or could I have an older version of the mthca module hanging
> around?
Not sure about any PP
This patch removes the last remaining typedef enums (minus bool
which /might/ justify it).
Signed-off-by: Tom Duffy <[EMAIL PROTECTED]>
Index: linux-kernel/dat-provider/dapl_cookie.c
===
--- linux-kernel/dat-provider/dapl_cookie.c
On Wed, Jun 15, 2005 at 04:30:47PM -0500, Troy Benjegerdes wrote:
> I'm getting a 'SDP: device mthca0 does nto support fast memory regions'
> error on PPC64 systems.. Is there something that needs to be done for
> PPC, or could I have an older version of the mthca module hanging
> around?
I'm no
On Thu, 2005-06-16 at 04:39, Itamar Rabenstein wrote:
> Hi Hal,
> I am trying to understand what is going here and i still dont see how this
> happan .
>
> This prints are only set in UP mode .(is this your system UP?)
Yes.
> the code is (function: dapl_evd_connection_callback):
> spin_lock_irq
Roland> Why is it useful to set the MTU before being able to
Roland> invoke ifconfig? The network interface can't do anything
Roland> useful until it is brought up.
Also, if this is useful for some application (I guess I can imagine
booting with NFS root), shouldn't the MTU setting be
Hal> [IPoIB] Add support for MTU module parameter. This is so
Hal> there can be a non default MTU at boot up if the
Hal> administrator so desires (prior to being able to invoke
Hal> ifconfig).
Why is it useful to set the MTU before being able to invoke ifconfig?
The network interfa
Itamar,
Can you seperate the RMR api change and hash table removal into a
seperate patch?
Merging the HCA and IA code into the dapl_ia.[ch] files doesn't feel
right to me. I think of the HCA as a DAT object just like IAs, EPs,
LMRs, etc, and hence it should have its own file.
james
On We
Initial changes to work with user AT.
Also, some other changes based on code inspection.
Signed-off-by: Hal Rosenstock <[EMAIL PROTECTED]>
Index: at.c
===
--- at.c(revision 2610)
+++ at.c(working copy)
@@ -118,7 +118,
お久しぶりです。はじめましての方もいらっしゃると思いますが改めまして
ご挨拶させ枦輿斌 侑?澱堙紬㉒輿謀㏄Ð 模崚 代表
藁科美智子でございます。
最初に申しておきますが、当クラブ「MAKE-UP
LOVE」の会報メールがご不要な
場合にはそのまま破棄してく枦輿斌 侑?澱堙紬㉒
尚、枦輿斌 侑?澱堙紬璽襪脇鹽戮版杰偁廚靴泙擦鵝£
それでは今回の緊急速報の内容に移りたいと思います。
この間のアンケートの結果が出ました。
お題は「今あなたが抱えている悩み」でした。
たくさんのご意見ありがとうございました。
そこでいくつか頂いたメールをご紹介したいと思います。
☆まずは東京都にお住まいの
As Itamar observed, the first message is printed out by
_raw_spin_lock() when the kernel is compiled for UP and spin lock
debugging.
dapl_ep_disconnect() is trying to obtain a lock that is already
locked. The message indicates that the lock was taken in
dapl_evd_connection_callback(). Ther
Title: RE: [openib-general] [RFC] [PATCH] SRQ API
> -Original Message-
> From: Caitlin Bestler [mailto:[EMAIL PROTECTED]]
> Sent: Wednesday, June 15, 2005 6:49 PM
> To: Roland Dreier
> Cc: openib-general@openib.org
> Subject: Re: [openib-general] [RFC] [PATCH] SRQ API
>
>
...
[IPoIB] Add support for MTU module parameter. This is so there can be a
non default MTU at boot up if the administrator so desires (prior to
being able to invoke ifconfig).
Signed-off-by: Hal Rosenstock <[EMAIL PROTECTED]>
Index: ipoib_main.c
==
Hi Hal,
I am trying to understand what is going here and i still dont see how this
happan .
This prints are only set in UP mode .(is this your system UP?)
the code is (function: dapl_evd_connection_callback):
spin_lock_irqsave(&ep->common.lock, ep->common.flags);
case on the event type
disconnect:
38 matches
Mail list logo