1. Feature Overview: a. Update bnx2x to 1.50.x version b. Feature Description This version includes new, more efficient FW, which improves performance. It also includes some minor performance improvements such as separate Rx and Tx MSI-X vectors 2. Feature Details: a. Architectures: 32-bit x86 64-bit Intel EM64T/AMD64 64-bit Itanium2 b. Upstream acceptance information: This version will be submitted to the upstream kernel in the next couple of weeks (by September 21st) 3. Business Justification: a. Why is this feature needed? Improved performance 4. Primary contact at Broadcom, email, phone Eilon Greenstein eilong +972-37552340
Thanks! Added to the list.
*** Bug 513232 has been marked as a duplicate of this bug. ***
*** Bug 511975 has been marked as a duplicate of this bug. ***
*** Bug 513935 has been marked as a duplicate of this bug. ***
*** Bug 516979 has been marked as a duplicate of this bug. ***
@Broadcom and other interested parties, We need to confirm that there is commitment to test for the resolution of this request during the RHEL 5.5 test phase, if it is accepted into the release. Please post a confirmation before Oct 16th, 2009, including the contact information for testing engineers.
10/13/2009 tungn Will do core test for this driver when image is ready.
------- Comment From lcm.com 2009-10-13 14:28 EDT------- IBM signing up to also provide testing feedback. Please coordinate through Peter Bogdanovic, pbogdano.
Hey Stanislaw, I had this bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=517378 for 5.5 asking to add bnx2x iscsi support. It turns out there is no bnx2i (no iscsi or scsi kernel bits) needed. Broadcom just needs some bnx2x bits. I wanted to see if you or we or broadcom (for storage drivers we have the vendors make patches but for net do you guys make the patches?) could make sure we get this brought in with the normal update in this bugzilla? Broadcom or I can give you a list of upstream commits, or Broadcom had even made some patches before if you want to go that route. Also we need this for RHEL6 too. Were you guys going to do any changes to bnx2x for RHEL 6?
(In reply to comment #9) > I wanted to see if you or we or broadcom (for storage drivers we have the > vendors make patches but for net do you guys make the patches?) For networking drivers usually vendors commits upstream and we do backports. > could make sure > we get this brought in with the normal update in this bugzilla? No problem. > Broadcom or I > can give you a list of upstream commits, or Broadcom had even made some patches > before if you want to go that route. Yes, you can provide patches, if I did not take them I can always use for compare to my backport. Are there any commits that are not in current 2.6.32 kernel (driver version 1.52.1)? I'm going to do full driver update - take all Broadcom upstream commits. > Also we need this for RHEL6 too. Were you guys going to do any changes to bnx2x > for RHEL 6? https://bugzilla.redhat.com/show_bug.cgi?id=464427
Change topic, in upstream we have no version 1.50.x, there is direct update from 1.48 -> 1.52 .
(In reply to comment #10) > Are there any commits that are not in current 2.6.32 > kernel (driver version 1.52.1)? I'm going to do full driver update - take all > Broadcom upstream commits. > The iSCSI patches are the same as RHEL6.0. Please include the following 4 bnx2x patches to support iSCSI: commit 993ac7b5183f82edc9696cd17faae03523e00e09 bnx2x: Add main CNIC interface functions. commit 37b091bacba7bd329eced9a56998b6247da414c4 bnx2x: Add hw init code to support iSCSI. commit e665bfda5b8fea586ddd028b26a6e2ed9e987c6b bnx2x: Refactor MAC address setup code. commit 2891290219d8a28a923560d53d24c00fa96fa09f bnx2x: Refactor bnx2x_sp_post(). In addition, these cnic patches below to support 10G iSCSI have dependency on the above bnx2x patches. Please include them as well. Thanks. commit c05e85a06e376f6b6d59e71e5333d707e956d78b cnic: Need to include net/ip6_checksum.h commit 71034ba845c9ff219373066f904286c0b7506922 cnic: Add main functions to support bnx2x devices. commit e251306556d30c9c678feca60204acaaa0784cbd cnic: Add bnx2x data structures. commit 86b53606ebec06b16be81c30fabdf8decc2be6b2 cnic: Refactor some code. commit 6053bbf7bbdbb2c94547f830ad07636c17d7024e cnic: Fix NETDEV_UP event processing. commit 6ef57a0ea77e7bf4123a06da895a0e26312b5737 cnic: Shutdown iSCSI ring during uio_close. commit 5e9b2dbfcbc9b180a0064d8a473a4652ee99c9ad cnic: Put uio init in separate function. commit ec0248ea35f6e1fc7426f377a7df071d997bdedd cnic: Put rx/tx ring allocation in separate function.
(In reply to comment #12) > In addition, these cnic patches below to support 10G iSCSI have dependency on > the above bnx2x patches. Please include them as well. Thanks. Is this only work that must to be done for bug 517378 ? Otherwise we need to have separate bug entry for that. If so, please create new bug entry (with dependency set to this one) and assign it to me. Thank you.
(In reply to comment #13) > > Is this only work that must to be done for bug 517378 ? Otherwise we need to > have separate bug entry for that. If so, please create new bug entry (with > dependency set to this one) and assign it to me. Thank you. There might be some cnic bug fixes later on. I'll create a "cnic bug fixes" BZ when necessary. Thanks.
Please note in https://bugzilla.redhat.com/show_bug.cgi?id=530121 IBM is reporting a problem with bnx2x and their HS22 blades and may require an additional patch or newer driver version to fix this.
This enhancement request was evaluated by the full Red Hat Enterprise Linux team for inclusion in a Red Hat Enterprise Linux minor release. As a result of this evaluation, Red Hat has tentatively approved inclusion of this feature in the next Red Hat Enterprise Linux Update minor release. While it is a goal to include this enhancement in the next minor release of Red Hat Enterprise Linux, the enhancement is not yet committed for inclusion in the next minor release pending the next phase of actual code integration and successful Red Hat and partner testing.
*** Bug 531882 has been marked as a duplicate of this bug. ***
Created attachment 369186 [details] 0001-bnx2x-update-to-1.52.1.patch
Created attachment 369187 [details] 0002-bnx2x-add-firmware-blob-5.0.21.0.patch
Patches not include iSCSI commits requested in comment #12. I'm going to work on them separate in 517378 bugzilla entry.
Eilon, In RHEL5.5 patch I include most commits up to: commit c458bc50e31631f7e9333ea6f1832fc54d4e2c1e Author: Eilon Greenstein <eilong> Date: Wed Aug 12 08:24:31 2009 +0000 bnx2x: update version to 1.52.1 All commits are listed inside patch. I do not backport below three commits, because in RHEL we have no MDIO support, hope this is ok: commit 01cd452846c98609dd3efbee0deea050e6706f02 Author: Eilon Greenstein <eilong> Date: Wed Aug 12 08:23:08 2009 +0000 bnx2x: MDC/MDIO CL45 IOCTLs commit f57a60256d02daba1316c98da472f02cd98a58d3 Author: Eilon Greenstein <eilong> Date: Wed Aug 12 08:23:11 2009 +0000 bnx2x: Supporting PHY FW upgrade commit cdea52128f6099e8f84459823c45790a78264022 Author: Eilon Greenstein <eilong> Date: Wed Aug 12 08:23:14 2009 +0000 bnx2x: Removing old PHY FW upgrade code Are we need some newer commits from net-next (other than mentioned in comment 12) ? Or maybe I should do full backport further up to 1.52.1-4 ?
Hi, The old MDC/MDIO code is no longer used by the utility that upgrades the PHY FW. How about adding the missing code at the end of the bnx2x_compat.h? I will send a patch with this suggestion. There is still an open issue with the Tx interrupt handling – a patch to move it back to NAPI is being prepared now and should be ready by Sunday. This one should be included as well – I will let you know once it is accepted. Thanks, Eilon
Created attachment 369202 [details] Add requied mdio functionality to bnx2x Adding the required mdio functionality into bnx2x_compat.h so the fw upgrade utility will be able to upgrade the PHY FW.
(In reply to comment #25) > The old MDC/MDIO code is no longer used by the utility that upgrades the PHY > FW. How about adding the missing code at the end of the bnx2x_compat.h? I will > send a patch with this suggestion. Great. > There is still an open issue with the Tx interrupt handling – a patch to move > it back to NAPI is being prepared now and should be ready by Sunday. This one > should be included as well – I will let you know once it is accepted. Ok, waiting for that fix. What about all other fixes currently in net-next ? I would like to include them in RHEL5, but there is also a FW upgrade change, so I'm not sure. Decision is yours :) Thanks!
Yes – having all of the net-next changes is preferred. Thanks, Eilon
Hi, Please update everything up to 9a4e328eb2bfa23b160558cff96e17ffa65ea5cf "bnx2x: Don't set netdev->trans_sta" in net-next (please note that this specific change is not applicable to RH5.5). The driver version is 1.52.1-5 (I'm also updating the BZ summary). Thanks, Eilon
Created attachment 370367 [details] 0001-bnx2x-update-to-1.52.1.patch
Created attachment 370368 [details] 0002-bnx2x-add-firmware-version-5.2.7.0.patch
Created attachment 370369 [details] 0003-bnx2x-add-MDIO-support.patch
Created attachment 370372 [details] 0004-bnx2x-update-to-1.52.1-5.patch
Brew build: https://brewweb.devel.redhat.com/taskinfo?taskID=2085280 Test packages (i686,x86_64,src) are available for public download here: http://people.redhat.com/sgruszka/bz515716/
Broadcom, please test and report back! thanks!
Created attachment 373041 [details] 0005-bnx2x-update-to-1.52.1-5__v2.patch This patch is the same as 0004-bnx2x-update-to-1.52.1-5.patch, but it's rebased atop Mike Christie patch for bug 516233, which also modifies cnic code.
in kernel-2.6.18-175.el5 You can download this test kernel from http://people.redhat.com/dzickus/el5 Please do NOT transition this bugzilla state to VERIFIED until our QE team has sent specific instructions indicating when to do so. However feel free to provide a comment indicating that this fix has been verified.
This patches ware not posted yet and are certainly not included in kernel-2.6.18-175.el5. I posted patch for bug 531882. Moving back to ASSIGNED.
I whould like to post patches (In reply to comment #34) > Test packages (i686,x86_64,src) are available for public download here: > http://people.redhat.com/sgruszka/bz515716/ Any partial testing results? I would like to hear everything is fine before positing patches to RKML :-)
TN tungn Tests cases passed: 520( IPV6 stress) 699 ( LED) 874 (Jumbo frame stress)1023 (Generic Bonding) Running loadunload. Will post result at the end of today or early tommorrow.
TN Test cases( Continued) Test cases executed: 1368 : loadunload ( Pass) 728 :continuous reboot test (pass)
Test Progress Below. If more information/details are required, please let me konw. Eddie. RH5.5 Inbox - Kernel 2.6.18-174 DUT: BCM95771x (common DUT) Linux - Everest 1.48.107 - 1.52.1 Passed Failed N/A Blocking Done 7 (33%) 0 (0%) 9 (42%) 0 (0%) 16 of 21 (76%)
in kernel-2.6.18-181.el5 You can download this test kernel from http://people.redhat.com/dzickus/el5 Please update the appropriate value in the Verified field (cf_verified) to indicate this fix has been successfully verified. Include a comment with verification details.
12/15/2009 TN : since this kernel (181) has the same version 1.52.1-5 (bnx2x) as kernel 174 that we have tested for the last 2 weeks, we only do sanity check on the new kernel ( 181). We are running 1368 loadunload only.Please post your concern if there is any. Thanks
12/16/2009 TN test case 1368 ( loadunload ) ran over night with more than 1000 iterations without any incident. Will update if there is any issue.
*** Bug 530121 has been marked as a duplicate of this bug. ***
In terms of test progress, BRCM PQA is planning on running first pass testing and regression tests by Beta snapshots. Current bnx2x first pass L2 testing (not including iSCSI offload) is at 77% complete. There is an issue found during automated iSCSI offload protocol testing. This issue is fixed in a newer version of bnx2x datapath fw. There are also additional iSCSI offload datapath fw fixes that were found during testing during our out of box testing on RH5.4. A separate bugzilla will be opened for these issues.
Technical note added. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. New Contents: = Broadcom Everest network devices = The bnx2x driver for Broadcom Everest network devices has been updated to version 1.52.1-5.(BZ#515716, BZ#522600)
*** Bug 516090 has been marked as a duplicate of this bug. ***
Test progress update - BRCM PQA current bnx2x first pass L2 testing (not including iSCSI offload) is at 90% complete.
------- Comment From coschult.com 2010-02-17 16:52 EDT------- Verified Beta 1 using Broadcom 57710 adapter. Driver present after booting and card detected and configured successfully. Tested using netperf (np3d test suite) on local network. All tests passed.
Event posted on 02-19-2010 11:07am EST by Glen Johnson ------- Comment From linuxram.com 2010-02-19 10:59 EDT------- This issue seems to be resolved on 5.5. We never encountered this problem in our 5.5 testing. Earlier we used to see this issue once in at least 3-4 reboots. Not anymore. This event sent from IssueTracker by jkachuck issue 355557
I have tested bnx2x 1.52.1-5 as provided in kernel 2.6.18-186.el5 x86_64 on a variety of Broadcom 57710/57711 based network adapters. I have not encountered any problems.
Hi, Broadcom PQA has moved to test 1.52.1-6 with the new updated fw. During the testing a new issue was found that during increasing netperf stress, the system scrolled message: bnx2x_start_xmit: 11208 Eth1 BUG! tx ring full when queue awake. A new bugzilla opened - Bug 567979 - bnx2x: net device is in XON state while the Tx ring is full Thanks, Gideon
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on therefore solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHSA-2010-0178.html