I saw no update here, but FYI it was promoted.
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1732892
Title:
[MIR] 18.04 rdma-core as replacement for older ibverbs code
To manage notifications about
There is a new dependency to one of the binaries in src:rdma-core.
Some of the DPDK libraries (in main) now depend on binary "rdma-core".
This part of src:rdma-core is the one that connects/enables dameons and module
loading configs which is why it is needed for some Mellanox based workloads.
Tha
Override component to main
ibverbs-providers 16.1-2 in bionic amd64: universe/net/optional/100% -> main
ibverbs-providers 16.1-2 in bionic arm64: universe/net/optional/100% -> main
ibverbs-providers 16.1-2 in bionic armhf: universe/net/optional/100% -> main
ibverbs-providers 16.1-2 in bionic i386:
Hi,
please note that need to remove the mlx4.conf from the kmod package.
Opened a bug on this issue LP: #1693503
yours,
Talat
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1732892
Title:
[MIR] 1
@Seth - ibverbs-providers is the new catch-all for libmlx[45] and a bit
more - so it is in after cyphermox is done with the current list.
There might be more dependencies as soon as we pick up rdma support in
more packages, but until I know details I can't request explicit binary
packages yet.
I
Thanks Benjamin, I'm embarrassed I didn't look in debian/changelog to
try to answer this question. Good job making it easily available :)
Of the packages listed, only librdmacm (CVE-2012-4516) has a CVE in our
database. The fix is nice and simple.
I also would have expected libmlx[45] -- is the s
The packages in component-mismatch are:
ibverbs-providers libibcm-dev libibcm1 libibumad-dev libibumad3
Unless you need something else that I haven't found, let's focus on just
bringing those binaries in main.
MIR approved.
** Changed in: rdma-core (Ubuntu)
Status: In Progress => Fix Com
I was asked to list the binaries that are to be accepted now that all acks seem
to be in.
The assumption here is that we want all that were already in main before, but
not e.g. utils.
So we check all that it builds and where it is today:
apt-cache policy $(awk '/^Package:/ {print $2}' debian/co
** Changed in: rdma-core (Ubuntu)
Assignee: Mathieu Trudel-Lapierre (cyphermox) => ChristianEhrhardt
(paelzer)
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1732892
Title:
[MIR] 18.04 rdma-cor
As Seth mentioned, this is a complex package.
Some notes:
- surprised to see infiniband/acm*.h shipped in icabm rather than a -dev
package
- rdma-core Recommends net-tools; if tools from that package are used in a
script or in doc, things should probably be changed to use ex. 'ip' commands
in
** Changed in: rdma-core (Ubuntu)
Status: Confirmed => In Progress
** Changed in: rdma-core (Ubuntu)
Assignee: (unassigned) => Mathieu Trudel-Lapierre (cyphermox)
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bug
Thanks for the review. To answer the question "Which source packages do
we get to delete as a result of this new package?", please look at the
first entry in debian/changelog:
* New upstream version. ibverbs-providers combines the source packages
libcxgb3, libipathverbs, libmlx4, libmlx5, li
Seth,
I can't think about such scenario for a number of reasons:
1. Such compiler/optimization should throw away the call to
ibv_cmd_get_context().
2. It should skip "write(context->cmd_fd, cmd, cmd_size) != cmd_size)" check
there.
Thanks
--
You received this bug notification because you are
Leon, by the way, are there any compilers, or compiler optimization
passes, that would make the same mistake that I did and assume the
"uninitialized resp" fields are really uninitialized, and write garbage
to the other structs? Or elide the assignments entirely?
Is there any way we can have confi
Hello, I reviewed rdma-core version 16.0-1 as checked into bionic. This
isn't a full security audit but rather a quick gauge of maintainability.
- CVE history is a bit hard to look for, since this source package is
quite new and composed of many previous source packages.
Which source packages
Could you please confirm that rdma-core will be included in Alpha-2 (Feb
1st)?
Thanks,
Noa
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1732892
Title:
[MIR] 18.04 rdma-core as replacement for olde
On Thu, Jan 11, 2018 at 07:10:06AM -, Leon Romanovsky wrote:
> You are missing the fact that rdma-core and RDMA kernel part are
> developed together and they are actually one piece.
It's even worse than that, I missed that the write() call in the userspace
component would have side-effects in
mlx4_init_contex:
https://github.com/linux-rdma/rdma-core/blob/master/providers/mlx4/mlx4.c#L149
-->
ibv_cmd_get_context:
https://github.com/linux-rdma/rdma-core/blob/master/libibverbs/cmd.c#L56
-->
.
-->
mlx4_ib_alloc_ucontext:
http://elixir.free-electrons.com/linux/latest/source/drivers/in
On Wed, Jan 10, 2018 at 08:21:48AM -, Leon Romanovsky wrote:
> For example, the resp.qp_tab_size is filled by kernel while calling to
> ibv_cmd_get_context, the communication is based on the fact that pointer
> to resp_v3.ibv_res is actually pointer to resp_v3.
Hello Leon, could you point out
Hi Seth,
The mlx4 driver is legacy one and the code in question was added in
2012.
For example, the resp.qp_tab_size is filled by kernel while calling to
ibv_cmd_get_context, the communication is based on the fact that pointer
to resp_v3.ibv_res is actually pointer to resp_v3.
In 2018, we would
On Tue, Jan 09, 2018 at 08:41:09AM -, Talat Batheesh wrote:
> Reading the report, there are a lot FALSE positives. For example, all
> resp.* warnings are false. resp is initialize to be zero.
Hello Talat, at least in the version I'm reviewing, this does not appear
to be the case.
Consider pr
> The rdma-core is checked with clang, gcc and sparse in Travis CI for
> every commit. We do invest time to clean legacy errors/warnings, but our
> resources are limited and it takes time to make it error/warning free.
Thank you so much Leon, that should give the sec evaluation some extra
confiden
Hi,
The rdma-core is checked with clang, gcc and sparse in Travis CI for
every commit. We do invest time to clean legacy errors/warnings, but our
resources are limited and it takes time to make it error/warning free.
Thanks
--
You received this bug notification because you are a member of Ubunt
On Tue, Jan 9, 2018 at 9:41 AM, Talat Batheesh
<1732...@bugs.launchpad.net> wrote:
[...]
I think everyone is aware that this is "the same in a new dress".
So one could just shove it through under the "same anyway" banner, but
why not taking the chance to make it better?
> please note that ther
Hi Seth,
Thank you for the review and for the detailed response.
The rdma-core was already accepted to debian and passed their acceptance
process.
In addition rdma-core was already integrated into the other major linux distors
(such as Fedora, RHEL, OpenSUSE, SLES, Debian ... ).
please note th
Hello Talat, I've started reviewing rdma-core a few days ago. My initial
thoughts are definitely mixed: on the one hand, if much of the code was
already in main in other packages, then just coalescing it into one
package is a low-risk change.
On the other hand, the code quality is all over the pla
Hi,
Any update with rdma-core ?
Yours,
Talat
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1732892
Title:
[MIR] 18.04 rdma-core as replacement for older ibverbs code
To manage notifications about
Status changed to 'Confirmed' because the bug affects multiple users.
** Changed in: rdma-core (Ubuntu)
Status: New => Confirmed
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1732892
Title:
** Changed in: rdma-core (Ubuntu)
Assignee: (unassigned) => Ubuntu Security Team (ubuntu-security)
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1732892
Title:
[MIR] 18.04 rdma-core as replacem
assigning the security team for a review
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/1732892
Title:
[MIR] 18.04 rdma-core as replacement for older ibverbs code
To manage notifications about this
The full transition of this should also include the Archive Team to
remove the old packages or files would conflict. I don't know how much
we follow Debian on this automatically as their AAs should have done the
same thing.
--
You received this bug notification because you are a member of Ubuntu
31 matches
Mail list logo
