Vlad,
Please pull this fix for bug 1609 from:
ssh://v...@sofa.openfabrics.org/~swise/scm/ofed-1.4.git ofed_1_4
Thanks,
Steve.
Original Message
Subject: [PATCH 2.6.30] xprtrdma: The frmr iova_start values are
truncated by the nfs rdma client.
Date: Fri, 24 Apr 2009 14:0
Steve,
You should have the same fix for the server side as well
Here is the patch
-vu
A bad cast causes the iova_start, which in this case is a DMA bus
address,
to be truncated on 32b systems. No cast is needed.
Signed-off-by: Steve Wise
---
net/sunrpc/xprtrdma/verbs.c |2 +-
1 fil
I'm seeing a server crash even with this frmr iova_start fix. I'm
debugging now...
Tziporet, I request we hold off on the new RC build until i resolve
these i386 issues.
Thanks,
Steve.
Vu Pham wrote:
Steve,
You should have the same fix for the server side as well
Here is the patch
-v
Vu Pham wrote:
Steve,
You should have the same fix for the server side as well
Here is the patch
Actually, I don't think this server change is needed. The server is
using a kernel virtual address as the iova_start, which is 32b on a 32b
system. The client, however, is using a bus address
Vu,
The server side is correct as-is. The crash I was seeing was due to
_removing_ the recast on the server side like you suggested. That cast
_is_ needed to avoid a sign extension when the void * is assigned into
the u64.
Vlad,
Please pull the client fix from:
ssh://v...@sofa.openfabri
Steve Wise wrote:
Vu,
The server side is correct as-is. The crash I was seeing was due to
_removing_ the recast on the server side like you suggested. That cast
_is_ needed to avoid a sign extension when the void * is assigned into
the u64.
Vlad,
Please pull the client fix from:
ssh://