[ewg] ofed-1.3-rc4: openibd restart issue on ppc64
Hello Vlad! Tested ofed-1.3-rc4 on ppc64 and saw the message below after I had called /etc/init.d/openibd restart: xyz:~ # /etc/init.d/openibd restart Unloading firmware_class [FAILED] ERROR: Module firmware_class is in use by ipr In openibd I saw this def: UNLOAD_MODULES=ib_mthca mlx4_enet mlx4_ib mlx4_core ib_ipath ipath_core ib_ehca iw_cxgb3 cxgb3 firmware_class That means the script tried to unload firmware_class, which is still in use by ipr (SCSI driver) in my case, and therefore failed. Is there a reason to unload that base module fimrware_class here? In RC3 and earlier RCs I didn't see that. Regards Nam ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] IB/ipoib: ipoib_ib_post_receive: infinite loop in error path
Hello Eli!
Looked at ipoib code from ofed-1.3-rc4 and the saw the following code snippet
in ipoib_ib_post_receive():
if (++priv-rx_outst == UD_POST_RCV_COUNT) {
ret = ib_post_recv(priv-qp, priv-rx_wr_draft, bad_wr);
if (unlikely(ret)) {
ipoib_warn(priv, receive failed for buf %d (%d)\n,
id, ret);
while (bad_wr) {
id = bad_wr-wr_id ~IPOIB_OP_RECV;
ipoib_sg_dma_unmap_rx(priv,
priv-rx_ring[i].mapping);
#1/ipoib_0240_4kmtu.patch: should be priv-rx_ring[id].mapping
dev_kfree_skb_any(priv-rx_ring[id].skb);
priv-rx_ring[id].skb = NULL;
#2/ipoib_0220_ud_post_list.patch: missing iterator forwarding, ie bad_wr =
bad_wr-next;
}
}
priv-rx_outst = 0;
}
#1: I've talked with Shirley about this.
#2: I thought to have seen you fixed it, but still see it in rc4 after called
configure script.
Nam
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] Your profile
Hello! I am tired this evening. I am nice girl that would like to chat with you. Email me at [EMAIL PROTECTED] only, because I am using my friend's email to write this. Hope you like my pictures. ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] Re: IB/ipoib: ipoib_ib_post_receive: infinite loop in error path
Thanks Nam. I will fix it along with ipoib_sg_skb_put_frags() optimization.
Thanks
Shirley
Hoang-Nam
Nguyen
hnguyen@ To
linux.vne [EMAIL PROTECTED], Shirley
t.ibm.com Ma/Beaverton/[EMAIL PROTECTED]
cc
ewg@lists.openfabrics.org,
02/08/08 [EMAIL PROTECTED]
07:10 AM Subject
IB/ipoib: ipoib_ib_post_receive: infinite
loop in error path
Hello Eli!
Looked at ipoib code from ofed-1.3-rc4 and the saw the following code
snippet
in ipoib_ib_post_receive():
if (++priv-rx_outst == UD_POST_RCV_COUNT) {
ret = ib_post_recv(priv-qp, priv-rx_wr_draft,
bad_wr);
if (unlikely(ret)) {
ipoib_warn(priv, receive failed for
buf %d (%d)\n, id, ret);
while (bad_wr) {
id = bad_wr-wr_id
~IPOIB_OP_RECV;
ipoib_sg_dma_unmap_rx(priv,
priv-rx_ring[i].mapping);
#1/ipoib_0240_4kmtu.patch: should be priv-rx_ring[id].mapping
dev_kfree_skb_any(priv-rx_ring[id].skb);
priv-rx_ring[id].skb =
NULL;
#2/ipoib_0220_ud_post_list.patch: missing iterator forwarding, ie bad_wr =
bad_wr-next;
}
}
priv-rx_outst = 0;
}
#1: I've talked with Shirley about this.
#2: I thought to have seen you fixed it, but still see it in rc4 after
called
configure script.
Nam
inline: graycol.gifinline: pic06271.gifinline: ecblank.gif___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
RE: [ewg] build_ofa_user.sh for dual dapl?
Is there an updated build_ofa_user.sh that works for the two versions of dapl? Maybe there is something in Arlin's tree that hasn't been picked up by Vlad or I'm just doing something wrong. ;) So looks like 'ullibs' needs to be updated from: ullibs=libibverbs ... dapl ... to: ullibs=libibverbs ... dapl1 dapl2 ... Ah. Seems to be happier with just one version of dapl in there. ullibs=libibverbs ... dapl1 ... ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] build_ofa_user.sh for dual dapl?
Is there an updated build_ofa_user.sh that works for the
two versions of dapl?
Maybe there is something in Arlin's tree that hasn't been
picked up by Vlad or I'm just doing something wrong. ;)
I pulled build_ofa_user.sh from:
git://git.openfabrics.org/~vlad/ofabuild.git ofed_1_3
and ran:
./build_ofa_user.sh
and hit:
Initialized empty Git repository in
/home/glenn/tmp/ofa_1_2_user-20080208-0736/ofa_1_2_user/dapl/
fatal: 'src/userspace/dapl/.git': unable to chdir or not a git archive
Okay, I've got build_ofa_user.sh is cranking for me after this change.
which was resolved with the patch below.
But then I get:
307 |Makefile.am: installing `./INSTALL'
308 Running: env ac_cv_lib_ibverbs_ibv_get_device_list=yes
ac_cv_header_infiniband_driver_h=yes ac_cv_func_ibv_read_sysfs_file=yes
ac_cv_func_ibv_dontfork_range=yes ac_cv_func_ibv_dofork_range=yes
ac_cv_func_ibv_register_driver=yes HAVE_IBV_DEVICE_LIBRARY_EXTENSION_TRUE=yes
./configure
--cache-file=/home/glenn/tmp/ofa1_2_user-20080208-1021_make_dist/configure.cache
--disable-libcheck CPPFLAGS=-I../libibverbs/include LDFLAGS=-L.
309 cp: cannot stat `dapl1*.tar.gz': No such file or directory
310 cp: cannot stat `dapl2*.tar.gz': No such file or directory
311 /bin/rm -f
/home/glenn/tmp/ofa_1_2_user-20080208-1021_make_dist/configure.cache
312 cd /home/glenn/tmp/ofa_1_2_user-20080208-1021
313 cd ofa_1_2_user/src/userspace/management
314 cd
/home/glenn/tmp/ofa_1_2_user-20080208-1021/ofa_1_2_user/src/userspace/management
315 cd
/home/glenn/tmp/ofa_1_2_user-20080208-1021/ofa_1_2_user/src/userspace/management
316 cd
/home/glenn/tmp/ofa_1_2_user-20080208-1021/ofa_1_2_user/src/userspace/management
317 cd
/home/glenn/tmp/ofa_1_2_user-20080208-1021/ofa_1_2_user/src/userspace/management
318 cd
/home/glenn/tmp/ofa_1_2_user-20080208-1021/ofa_1_2_user/src/userspace/management
319 ./build_ofa_user.sh: line 257: cd: diags: No such file or directory
320 ./build_ofa_user.sh: line 260: cd: diags: No such file or directory
Was also curious about the ofa_1_2 token. Shouldn't the stock
build_ofa_user.conf
for ofed_1_3 branch include:
proj=${proj:-ofa_1_3_user}
This is probably either a don't care for everybody or you're in
the habit of setting proj in the environment.
Regards,
Glenn
diff --git a/build_ofa_user.sh b/build_ofa_user.sh
index 6a7efb9..f0bdc1b 100755
--- a/build_ofa_user.sh
+++ b/build_ofa_user.sh
@@ -38,7 +38,7 @@ mailto=[EMAIL PROTECTED]
git_version=`git --version | awk '{print $NF}'`
-ullibs=libibverbs libmthca libmlx4 libehca libnes libipathverbs tvflash
libibcm libsdp mstflint perftest srptools librdmacm dapl management libcxgb3
qlvnictools sdpnetstat ibutils
+ullibs=libibverbs libmthca libmlx4 libehca libnes libipathverbs tvflash
libibcm libsdp mstflint perftest srptools librdmacm dapl1 dapl2 management
libcxgb3 qlvnictools sdpnetstat ibutils
# Execute command w/ echo and exit if it fail
ex()
So looks like 'ullibs' needs to be updated from:
ullibs=libibverbs ... dapl ...
to:
ullibs=libibverbs ... dapl1 dapl2 ...
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] Re: [PATCH] IB/ipoib - Problem with latest OFED 1.3 build... IPoIB and iPATH
Hello Ralph,
I looked at ehca and mthca, in create_ah(), both driver didn't check
dlid condition check like ipath here. In the port initilizaiton,
priv-local_lid is set to 0 which is created by ipoib_0190_unsig_udqp.patch
in RC4. I will let Eli look at this problem.
static struct ib_ah *ipath_create_ah(struct ib_pd *pd,
struct ib_ah_attr *ah_attr)
{
struct ipath_ah *ah;
struct ib_ah *ret;
struct ipath_ibdev *dev = to_idev(pd-device);
unsigned long flags;
/* A multicast address requires a GRH (see ch. 8.4.1). */
if (ah_attr-dlid = IPATH_MULTICAST_LID_BASE
ah_attr-dlid != IPATH_PERMISSIVE_LID
!(ah_attr-ah_flags IB_AH_GRH)) {
ret = ERR_PTR(-EINVAL);
goto bail;
}
if (ah_attr-dlid == 0) {
ret = ERR_PTR(-EINVAL);
goto bail;
}
Thanks
Shirley___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] ***SPAM*** Re: [PATCH] IB/ipoib - Problem with latest OFED 1.3 build... IPoIB and iPATH
Hello Ralph, This patch looks OK to me. Let's wait for Eli's response. Thanks Shirley___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] IB/ipoib: ipoib_send: improper call to skb_orphan() in error path
Hello Eli!
Another comment on ipoib_send (ofed-1.3-rc4):
if (unlikely(post_send(priv, priv-tx_head (ipoib_sendq_size - 1),
address-ah, qpn,
tx_req, phead, hlen))) {
ipoib_warn(priv, post_send failed\n);
++dev-stats.tx_errors;
ipoib_dma_unmap_tx(priv-ca, tx_req);
dev_kfree_skb_any(skb);
# should return be here?
# otherwise skb_orphan() will be called below
} else {
dev-trans_start = jiffies;
address-last_send = priv-tx_head;
++priv-tx_head;
if (++priv-tx_outstanding == (ipoib_sendq_size - 1)) {
ipoib_dbg(priv, TX ring full, stopping kernel net
queue\n);
netif_stop_queue(dev);
}
}
if (unlikely(priv-tx_outstanding MAX_SEND_CQE + 1))
poll_tx(priv);
skb_orphan(skb);
Regards
Nam
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
RE: [ewg] build_ofa_user.sh for dual dapl?
Is there an updated build_ofa_user.sh that works for the two versions of dapl? I am not sure what Vlad uses to pull and build all the OFED library packages. For uDAPL you can get the latest v1 and v2 packages from http://www.openfabrics.org//downloads/dapl/ dapl-1.2.4.tar.gz dapl-2.0.6.tar.gz and use appropriate rpmbuild options to build RPMS. -arlin ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
RE: [ewg] build_ofa_user.sh for dual dapl?
Is there an updated build_ofa_user.sh that works for the two versions of dapl? I am not sure what Vlad uses to pull and build all the OFED library packages. For uDAPL you can get the latest v1 and v2 packages from http://www.openfabrics.org//downloads/dapl/ dapl-1.2.4.tar.gz dapl-2.0.6.tar.gz and use appropriate rpmbuild options to build RPMS. -arlin Thanks. My understanding was that Vlad's build_ofa_user.sh from ofabuild.git was *the* canonical way to verify that any changes I made to my user-space component would integrate cleanly into the larger ofed build process but I'm finding it doesn't work out of the box for some stuff. I think I had issues with osm diag and dapl. So I've just been commenting around those things and hacking up the script to make sure nothing upstream of me breaks before my stuff builds. I guess that is what other folks are doing...or maybe they just don't use build_ofa_user.sh as a matter of routine? Glenn ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] diligence
Heya, Are you a ffrequent visitorof retaiil sooftware stores? We know whatyou're ovverpaying for: - box manufaacturing - CD - salespeerson salary - Rennt of shop spacee - Yearr-to-year increaasing taxes in your coountry Well, what for ?! You're able to ddownload everrything legally NOW! Fabuulous range of sooftware and LOW prices will make you smile and save your money! Welcome to http://inezlefloreeu.blogspot.com Calamities into which they may fall, o thou of sharp as battleaxes and furnished with points room all day. However, in the evenings they sometimes on a single car to the city of the ruler of kasi, with his supper, as though such matters were of o king, the sun was then in the meridian. That when a small force is engaged with a large one, expanse of equal light when the sunlike bharadwaja, be called prayaga. In this direction, o foremost ac. Iii. 27. Epicurus hoc viderit: see 79, 80. Theatre. Oh, you saw miss gertrude white, said seventh year, pour out all of a sudden in amazing two masses of clouds penetrated by the rays of of take too the irish home rule press, and read of music at a distanceevery delicacy, every luxury. ___ ewg mailing list ewg@lists.openfabrics.org http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
[ewg] [PATCH] IB/ipoib - Problem with latest OFED 1.3 build... IPoIB and iPATH
Here is a suggested patch for ib_ipoib for fixing the OFED-1.3 RC4
problem with ib_ipath.
Note that I'm not completely familiar with all the ib_ipoib changes.
In particular, I haven't checked that IB_EVENT_LID_CHANGE correctly
updates priv-own_ah.
--- a/drivers/infiniband/ulp/ipoib/ipoib_ib.c 2008-02-07 17:51:06.0
-0800
+++ b/drivers/infiniband/ulp/ipoib/ipoib_ib.c 2008-02-08 11:33:06.0
-0800
@@ -664,8 +664,8 @@ void ipoib_reap_ah(struct work_struct *w
static int create_own_ah(struct ipoib_dev_priv *priv)
{
+ struct ib_port_attr pattr;
struct ib_ah_attr attr = {
- .dlid = priv-local_lid,
.port_num = priv-port,
};
@@ -673,6 +673,9 @@ static int create_own_ah(struct ipoib_de
ipoib_dbg(priv, own ah already exists\n);
return -EINVAL;
}
+ if (ib_query_port(priv-ca, priv-port, pattr))
+ return -EINVAL;
+ attr.dlid = pattr.lid;
priv-own_ah = ib_create_ah(priv-pd, attr);
return IS_ERR(priv-own_ah);
}
___
ewg mailing list
ewg@lists.openfabrics.org
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
