The issue is that current deb packages are built with autotools, and
apparently clang is required for asan now:
https://llvm.org/bugs/show_bug.cgi?id=22757
I am experimenting with building with "AUTOCONF_BUILD=no" in debian/rules,
so far I had to remove all but the BugPoint tests from
Subject: unblock: librdmacm/1.0.15-1+deb7u1
Package: release.debian.org
User: release.debian@packages.debian.org
Usertags: unblock
Severity: normal
Please unblock package librdmacm
The attached debdiff includes the fix for CVE-2012-4516.
unblock librdmacm/1.0.15-1+deb7u1
diff -Nru librdmacm-1.0.15/debian/.pc/.quilt_patches
librdmacm-1.0.15/debian/.pc/.quilt_patches
[...]
diff -Nru librdmacm-1.0.15/debian/.pc/.quilt_series
librdmacm-1.0.15/debian/.pc/.quilt_series
[...]
diff -Nru librdmacm-1.0.15/debian/.pc/.version
Whilst this has now been fixed in unstable, it was via the upload of a
new upstream which adds over 2000 lines of new code (and doesn't even
directly include the security fix) and with a debhelper compat bump
thrown in on the packaging side.
It's possible someone might feel inclined
The first vulnerable version in Debian is 1.0.14. Upstream introduced
ACM support (where the vulnerability exists) in version 1.0.12, so
Debian's 1.0.10 is not vulnerable.
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact
Package: libibumad
Version: 1.2.3-20090314-1.1
Severity: normal
Tags: patch pending
Dear maintainer,
I've prepared an NMU for libibumad (versioned as 1.3.8-0.1) and
uploaded it to DELAYED/10. Please feel free to tell me if I
should delay it longer.
Regards.
libibumad_1.3.8-0.1.debian.tar.gz
Package: libibmad
Version: 1.2.3-20090314-1.1
Severity: normal
Tags: patch pending
Dear maintainer,
I've prepared an NMU for libibmad (versioned as 1.3.9-0.1) and
uploaded it to DELAYED/10. Please feel free to tell me if I
should delay it longer.
Regards.
libibmad_1.3.9-0.1.debian.tar.gz
Package: infiniband-diags
Version: 1.4.4-20090314-1.1
Severity: normal
Tags: patch pending
Dear maintainer,
I've prepared an NMU for infiniband-diags (versioned as 1.6.1-0.1) and
uploaded it to DELAYED/10. Please feel free to tell me if I
should delay it longer.
Regards.
Package: ibutils
Version: 1.2-OFED-1.4.2-1.3
Severity: normal
Tags: patch pending
Dear maintainer,
I've prepared an NMU for ibutils (versioned as 1.5.7-0.1) and
uploaded it to DELAYED/10. Please feel free to tell me if I
should delay it longer.
Regards.
ibutils_1.5.7-0.1.debian.tar.gz
Package: opensm
Version: 3.2.6-20090317-2.1
Severity: normal
Tags: patch pending
Dear maintainer,
I've prepared an NMU for opensm (versioned as 3.3.15-0.1) and
uploaded it to DELAYED/10. Please feel free to tell me if I
should delay it longer.
Regards.
opensm_3.3.15-0.1.debian.tar.gz
Package: wnpp
Owner: Roland Dreier r...@debian.org
Severity: wishlist
* Package name: libnes
Version : 1.1.3
Upstream Author : Intel Corporation
* URL : http://www.openfabrics.org/
* License : GPLv2 or BSD
Programming Lang: C
Description : Userspace
I just tried building the Debian source of the openmpi 1.4.3-2.1 package
in pbuilder using the latest libibvers 1.1.5-1, and the build completed
successfully. It's going to be hard to figure out what went wrong in
your setup without more specific information.
Is it possible that you configured
From Debian Policy 10.2:
Packages that use libtool to create and install their shared
libraries install a file containing additional metadata (ending in
.la) alongside the library. For public libraries intended for use
by other packages, these files normally should not be
I ran into the same bug, packages coming from:
http://pkg-ofed.alioth.debian.org/howto/infiniband-howto-2.html#ss2.1
So you added a non-default package repository? Sounds like the bug is
in the packages from there.
However, I can't get a good build (to workaround):
gcc -DHAVE_CONFIG_H
There really should be no way for the ibv_cmd_reg_xrc_rcv_qp symbol to
get into binaries from a real Debian archive, since the package I
uploaded has nothing XRC-related.
Is everyone having this bug absolutely sure that the problem came from a
libmlx4 package from a real debian repository? If so
Package: wnpp
Owner: Roland Dreier rol...@digitalvampire.org
Severity: wishlist
* Package name: libcxgb4
Version : 1.1.1
Upstream Author : Chelsio Communications, Inc.
* URL : http://www.openfabrics.org/
* License : GPLv2 or BSD
Programming Lang: C
)
From e51c7b1ab05d4a6fe4d153b2769290d37e077479 Mon Sep 17 00:00:00 2001
From: Roland Dreier rol...@purestorage.com
Date: Thu, 03 Mar 2011 16:39:35 -0800
Subject: [PATCH] vnc: tight: Fix crash after 2GB of output
If one leaves a VNC session with tight compression running for long
enough, Qemu crashes
Philippe, I believe your problems are a separate bug (which I think is a
bad interaction between the qemu VGA BIOS and grub2).
--
To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org
with a subject of unsubscribe. Trouble? Contact listmas...@lists.debian.org
Package: qemu-kvm
Version: 0.14.0+dfsg-1~tls
Severity: important
I'm running a 32-bit guest (Xubuntu 10.04 to be precise) on a
64-bit host with the command line:
kvm -drive file=xubuntu.img,if=virtio,format=qcow2,boot=on -net
nic,model=virtio -net user -m 1024 -vga vmware -usbdevice tablet
Advocates:
http://lists.debian.org/debian-newmaint/2011/02/msg00019.html
Comment: Add Roland Dreier rol...@digitalvampire.org as a Debian Maintainer
Date: Thu, 17 Feb 2011 09:38:24 -0800
Action: import
Data:
-BEGIN PGP PUBLIC KEY BLOCK-
Version: GnuPG v1.4.10 (GNU/Linux
Package: memtest86+
Version: 4.00-2.1
Severity: normal
Tags: patch
*** Please type your report below this line ***
I have a system (unreleased big server) where the BIOS reports more
than 32 e820 entries to describe the RAM. memtest86+ has a hard-coded
limit of 32 e820 entries, so it crashes on
On the other hand, if that package is only ever useful on Linux, it
probably could be added to the P-a-s database (Packages arch specific)
and dropped from libopenmpi-dev's dependencies for non-Linux archs.
As far as I know, the FreeBSD kernel has no support for userspace access
to
I've found the following:
http://sources.freebsd.org/HEAD/src/sys/contrib/rdma/rdma_device.c
So there might be some effort to support that, but that's not here yet.
Shall I close this bug, and request the P-a-s addition? That would be
the attached patch. We also got openmpi fixed in
Package: kvm
Version: 79+dfsg-2
Severity: wishlist
Tags: patch
Hi, given that kvm is still undergoing rapid development, I often find
myself debugging various issues, and it is useful to have debugging
symbols for the userspace component available, if only to report sane
backtraces upstream.
(The Duplicated entry error seems to be an unrelated buglet caused by
grub-probe not discarding the previous map[] contents when regenerating
device.map)
Ugh. That deserves a new bug I think. Could you file it as a reminder?
OK, will file later.
Ok, here's a complete patch.
Next time, please don't add stuff to another bug just because it's similar.
File a new one instead.
Sorry -- I thought it was basically the same bug (unknown device name
pattern for root filesystem causes grub to destroy device.map).
Could you use the attached patch on grub-common, and
[ARGH!! sent to the wrong bug, sorry!!]
Next time, please don't add stuff to another bug just because it's similar.
File a new one instead.
Sorry -- I thought it was basically the same bug (unknown device name
pattern for root filesystem causes grub to destroy device.map).
Could you use
I have a similar problem with HP DL380 servers with a RAID controller
that uses the cciss driver. Disks show up under /dev/cciss like:
/dev/cciss/c0d0p6 on / type ext3 (rw,errors=remount-ro)
And since grub no longer understands that, grub-mkdevicemap leaves me
with an empty device.map and I
Package: wnpp
Owner: Roland Dreier [EMAIL PROTECTED]
Severity: wishlist
* Package name: libcxgb3
Version : 1.1.5
Upstream Author : Chelsio Communications, Inc.
* URL : http://www.openfabrics.org/
* License : GPLv2 or BSD
Programming Lang: C
Description
Package: wnpp
Owner: Roland Dreier [EMAIL PROTECTED]
Severity: wishlist
* Package name: libipathverbs
Version : 1.1
Upstream Author : QLogic Corp. [EMAIL PROTECTED]
* URL : http://www.openfabrics.org/
* License : GPLv2 or BSD
Programming Lang: C
Description
Package: wnpp
Owner: Roland Dreier [EMAIL PROTECTED]
Severity: wishlist
* Package name: librdmacm
Version : 1.0.5
Upstream Author : Sean Hefty [EMAIL PROTECTED]
* URL : http://www.openfabrics.org/
* License : GPLv2 or BSD
Programming Lang: C
Description
Package: wnpp
Owner: Roland Dreier [EMAIL PROTECTED]
Severity: wishlist
* Package name: libmlx4
Version : 1.0
Upstream Author : Roland Dreier [EMAIL PROTECTED]
* URL : http://www.openfabrics.org/
* License : GPLv2 or BSD
Programming Lang: C
Description
Upstream is using the patch in the mail linked below to fix this issue:
http://lists.gnu.org/archive/html/bug-automake/2007-08/msg00010.html
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Package: automake
Version: 1:1.10+nogfdl-1
When using automake to generate a non-recursive Makefile for a package that has
multiple subdirectories, 'make clean' does not delete all generate files. This
causes 'make distcheck' to fail. I am attaching a complete example package that
shows this
I believe this is caused by a bug in autotools, which fails to clean
generated files properly if a package uses non-recursive make and has
multiple subdirectories. I have filed a trimmed-down test case to
reproduce this as an automake bug in bug#454336.
- Roland
--
To UNSUBSCRIBE, email to
Package: udev
Version: 0.114-2
Severity: normal
Tags: patch
Unfortunately the patch that I included in the report for bug #427993 that
added handling of the rdma_ucm module's misc device had a typo in the rules.
Although the module name is rdma_ucm, the device that is created has
Package: udev
Version: 0.114-2
Severity: normal
Tags: patch
Unfortunately the patch that I included in the report for bug #427993 that
added handling of the rdma_ucm module's misc device had a typo in the rules.
Although the module name is rdma_ucm, the device that is created has
Package: openmpi
Version: 1.2.4-3
Severity: minor
The openmpi package has Build-Depends: libsysfs-dev for some (probably
historical) reason. However, I have just tested a build on a system
without libsysfs installed at all, and it completed fine.
The upstream changelog says:
- Removed
Any chance of getting a new version of kvm uploaded soon? kvm 28 is out
already, so the kvm 18 that's in the archive is 10 releases out of date.
And kvm 18 won't work with the kvm module that's in any kernel (it's too new
for 2.6.21 and too old for 2.6.22).
Thanks...
Package: udev
Version: 0.105-4
Severity: wishlist
Tags: patch
*** Please type your report below this line ***
The rdma_ucm module creates a misc char device:
KERNEL==rdma_cm
SUBSYSTEM==misc
DRIVER==
ATTR{abi_version}==4
ATTR{dev}==10:62
The /dev node for this file should be
libibverbs: Fatal: couldn't read uverbs ABI version.
This is the libibverbs message. For future versions I will look at
making this more informative and also providing a way to silence it
completely.
--
[0,0,0]:
As part of the weekend bug squashing party, I've sponsored this upload
for you since your usual sponsor is unresponsive.
Thanks!
I had sent 3 emails to my usual sponsor with no response :(
(And this is my second sponsor after my first sponsor told me he was
too busy to handle my packages...)
I'm not sure what is being asked for here. Would you like libibverbs1
to work if sysfs is not mounted, or are you just looking for a better
error message that explains the cause of the failure better?
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact
I'm not sure what is being asked for here. Would you like libibverbs1
to work if sysfs is not mounted, or are you just looking for a better
error message that explains the cause of the failure better?
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact
sorry, that last email was meant for 418014, not this bug...
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
A 1.5.0.1 maintenance release has just come out with a few fixes, so I
think it would definitely be appropriate to upload packages of the
1.5.0 series now.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
I'm confused -- what is the current status? Do you understand why
mirror/http/proxy should not apply when fetching preseed/include and
preseed/run files, or do you need more clarification?
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL
Package: debian-installer
Version: 20061102
I'm trying to use d-i to netboot a system and run a preseeded install,
with the preseed config being loaded via http. The configuration of
my network is such that I have to use an http proxy to reach Debian
mirrors, so I have a mirror/http/proxy
Sorry -- I have a couple of pending uploads that will fix this, but I
wanted to get a few other things fixed as well.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
How about the following:
Description: A userspace driver for Mellanox InfiniBand HCAs
libmthca is a device-specific driver for Mellanox InfiniBand host
channel adapters (HCAs) for the libibverbs library. This allows
userspace processes to access Mellanox HCA hardware directly with
low
This is actually a bug in the infiniband/arch.h shipped with
libibverbs-dev. I will fix this in the next upload of libibverbs.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Package: udev
Version: 0.070-2
Severity: wishlist
*** Please type your report below this line ***
Please change the udev rules for ucm and uverbs to add a group:
KERNEL=uverbs*, NAME=infiniband/%k, MODE=0660, GROUP=rdma
KERNEL=ucm*, NAME=infiniband/%k, MODE=0660, GROUP=rdma
As you can
On Oct 10, Roland Dreier [EMAIL PROTECTED] wrote:
I suggest the name rdma for the group name because these device
nodes give userspace programs access to RDMA capabilities from
InfiniBand and (someday) iWARP hardware.
Of course, this requires the creation of the rdma group. I'm
I have issued an ITP but I am still working through some review
comments on the libibverbs package.
OK, if you are going to maintain the other relevant package then I can
just use the name you will decide.
Thanks, I added postinst/postrm scripts to my libibverbs package that
add/remove
Package: wnpp
Severity: wishlist
Owner: Roland Dreier [EMAIL PROTECTED]
* Package name: libibverbs
Version : 1.0-rc1
Upstream Author : Roland Dreier [EMAIL PROTECTED]
* URL : http://www.openib.org/
* License : Dual GPL or BSD
Description : A library
Package: wnpp
Severity: wishlist
Owner: Roland Dreier [EMAIL PROTECTED]
* Package name: libmthca
Version : 1.0-rc1
Upstream Author : Roland Dreier [EMAIL PROTECTED]
* URL : http://www.openib.org/
* License : Dual GPL and BSD
Description : A userspace
Package: xemacs21
Version: 21.4.16-1
Severity: normal
Tags: patch
I read my email with Gnus. With the version included with xemacs21, if I get
email from a sender with non-ASCII characters in his or her name, I get an error
like the following when trying to enter the group:
Fetching headers
Package: libdbus-cil
Version: 0.22+cvs.200412122210-0ubuntu1
Severity: important
Justification: fails to build from source
Just tried to build dbus-mono. The first time through the build failed
almost immediately because gacutil was not available (because mono-gac
wasn't installed). It seems
Package: oprofile
Severity: important
oprofile depends on oprofile-modules0.8, but oprofile-modules0.8 packages only
exist
for 2.4 kernels (2.6 debian kernel-image packages already contain oprofile.ko,
and
it's quite easy for users to build oprofile into 2.6 packages with make-kpkg).
Al Not being a user of aptitude, I hadn't seen the behavior you
Al described. Was this just a fluke, or something I can adjust
Al in the package to improve it?
Good question, I'm not really an aptitude expert either. Maybe the
recommends: could become suggests: but that's possibly
Greg glibc issue Try upgrading it to a newer version.
Looks like it. The udev tests behave as you say, and even simpler,
the following program prints 0x12c00 but creates /dev/infiniband/xxx
with major 0x2c. This is with Debian libc6 2.3.2.ds1-20, which is the
newest available in Debian. Do
(I guess this bug should be reassigned to libc6)
It seems that the change to glibc's mknod was applied about 15 months
ago, between glibc 2.3.2 and glibc 2.3.3. Here's the patch taken from
glibc CVS:
Index: glibc-2.3.2/sysdeps/unix/sysv/linux/xmknod.c
62 matches
Mail list logo
