Description of problem: I an running Mageia linux on a dell poweredge 1950, with a broadcom ethernet card (Broadcom Corporation NetXtreme II BCM5708 Gigabit Ethernet ) It uses the bnx2 driver, which is provided by the kernel-firmware-nonfree package. The server provides http (apache) and ftp (vsftpd) services. with Mageia 2, 3, 4, I had no problems, but with Mageia 5, some users reports network access problems, on ftp and http. I had a tcpdump, I can see many ack dup and retransmission packets. I then reboot the mageia 5 with an old mageia 4 kernel (3.14.41-server-1.mga4) and all problems are solved. Version-Release number of selected component (if applicable): kernel-server-latest-3.19.8-3.mga5 kernel-server-3.19.8-3.mga5-1-1.mga5 kernel-server-3.14.41-1.mga4-1-1.mga4 kernel-firmware-20150309-1.mga5 kernel-firmware-nonfree-20150510-1.mga5.nonfree How reproducible: boot on kernel-server-3.19.8-3.mga5-1-1.mga5 Steps to Reproduce: 1. 2. 3. Reproducible: Steps to Reproduce:
Thanks for the report. Assigning to our kernel maintainer. You might want to test the 4.1 kernel that is currently in Core/Updates Testing to see if it fixes your issue.
Assignee: bugsquad => tmb
I have also tried the kernel-tmb-desktop-latest-3.19.8-1.mga5 package : same problem I can't find any 4.1 kernel on update_testing (http://distrib-coffee.ipsl.jussieu.fr/pub/linux/Mageia/distrib/5/x86_64/media/core/updates_testing/). So I installed the 4.1 kernel from cauldron : kernel-server-4.1.6-5.mga6-1-1.mga6 kernel-firmware-nonfree-20150722-1.mga5.nonfree kernel-firmware-20150722-1.mga5 and it seems to work. I will wait a day to confirm (or not)
the network is better with kernel 4.1.6, but now the server hangs with the following message : Sep 16 16:53:50 esurfmar.meteo.fr kernel: INFO: rcu_sched detected stalls on CPUs/tasks: { 0} (detected by 2, t=6002 jiffies, g=94332, c=943 Sep 16 16:53:50 esurfmar.meteo.fr kernel: Task dump for CPU 0: Sep 16 16:53:50 esurfmar.meteo.fr kernel: swapper/0 R running task 0 0 0 0x00200008 Sep 16 16:53:50 esurfmar.meteo.fr kernel: 0000000000000000 0000000000000000 0000000000000000 ffffffff81c04000 Sep 16 16:53:50 esurfmar.meteo.fr kernel: 0000000000000000 ffffffffffffff10 ffffffff8101eb2c 0000000000000010 Sep 16 16:53:50 esurfmar.meteo.fr kernel: 0000000000000246 ffffffff81c03ea8 0000000000000018 0000000000000000 Sep 16 16:53:50 esurfmar.meteo.fr kernel: Call Trace: Sep 16 16:53:50 esurfmar.meteo.fr kernel: [<ffffffff8101eb2c>] ? mwait_idle+0x8c/0x150 Sep 16 16:53:50 esurfmar.meteo.fr kernel: [<ffffffff8101f49f>] ? arch_cpu_idle+0xf/0x20 Sep 16 16:53:50 esurfmar.meteo.fr kernel: [<ffffffff810b2af7>] ? cpu_startup_entry+0x327/0x3a0 Sep 16 16:53:50 esurfmar.meteo.fr kernel: [<ffffffff816ef59c>] ? rest_init+0x7c/0x80 Sep 16 16:53:50 esurfmar.meteo.fr kernel: [<ffffffff81d31f7a>] ? start_kernel+0x42d/0x43a Sep 16 16:53:50 esurfmar.meteo.fr kernel: [<ffffffff81d31120>] ? early_idt_handler_array+0x120/0x120 Sep 16 16:53:50 esurfmar.meteo.fr kernel: [<ffffffff81d31339>] ? x86_64_start_reservations+0x2a/0x2c Sep 16 16:53:50 esurfmar.meteo.fr kernel: [<ffffffff81d31476>] ? x86_64_start_kernel+0x13b/0x14a
Yep, that's one of the reasons I pulled 4.1.6 from testing... I'm working on a 4.1.7 that should land on mirrors tonight /tomorrow
There is now a 4.17 in testing to try out
Depends on: (none) => 16655
I have installed kernel-server-4.1.7-1.mga5-1-1.mga5 on Thu 17 Sep 2015 and all seems fine : - no more network problem - no more rcu problems thanks for your help !
Mass-reassigning all bugs with "kernel" in the Source RPM field that are assigned to tmb, to the kernel packagers group, because tmb is currently MIA.
Assignee: tmb => kernel
(In reply to eric gerbier from comment #6) > I have installed kernel-server-4.1.7-1.mga5-1-1.mga5 on Thu 17 Sep 2015 > and all seems fine : > - no more network problem > - no more rcu problems > > thanks for your help ! Thanks for the feedback :-) Closing
Status: NEW => RESOLVEDCC: (none) => marja11Resolution: (none) => FIXED