hp-health/hpasmlited (Debian) - segfault

[Tom__PL]

Hi

My server is:

• ProLiant DL180 Gen9

Firmware:

• iLO ver. 2.44
• System ROM ver. U20 v2.20
• Smart Array P440 ver. 4.02

System:

• Debian 8.6 (Jessie)
• Linux Kernel 4.7.8
• hp-health ver.  10.40-1815.49

Syndrom:

When hp-health is started I got hpasmlited segfault error:

hpasmlited -f /dev/hpilo

kernel: x86/PAT: hpasmlited:26867 map pfn expected mapping type uncached-minus for [mem 0x7916c000-0x7916dfff], got write-back

kernel: hpasmlited[26867]: segfault at 0 ip 000000000040605b sp 00007ffe43d62f10 error 4 in hpasmlited[400000+3a000]

a closer look with strace tool:

mknod("/dev/indc", S_IFCHR|0644, makedev(255, 184)) = 0
stat("/sys/firmware/efi", 0x7ffc41c8b180) = -1 ENOENT (No such file or directory)
stat("/proc/efi", 0x7ffc41c8b180)       = -1 ENOENT (No such file or directory)
iopl(0x3)                               = 0
rt_sigprocmask(SIG_BLOCK, ~[HUP TERM RTMIN RT_1], NULL, 8) = 0
rt_sigaction(SIGTERM, {0x402a70, [], SA_RESTORER|SA_RESTART, 0x7f7e098e18d0}, NULL, 8) = 0
rt_sigaction(SIGHUP, {0x402a70, [], SA_RESTORER|SA_RESTART, 0x7f7e098e18d0}, NULL, 8) = 0
clone(child_stack=0, flags=CLONE_CHILD_CLEARTID|CLONE_CHILD_SETTID|SIGCHLD, child_tidptr=0x7f7e0a1099d0) = 4081
exit_group(0)                           = ?
+++ exited with 0 +++

I think that problem may be related with UEFI disabled - my server runs with Legacy Boot.

Best Regards.

Tom.

[Jimmy Vance]

try putting "nopat" on the hpasmlited command line

[Tom__PL]

Putting "nopat" really helped, but on the kernel boot arguments (not on the hpasmlited command line).  :)

I think this is temporary workaround because kernel with PAT disabled works with degradated performance.

Regards.

[Exploitation67]

+1

We have this problem too on our server.

Putting "nopat" on the kernel boot arguments helped.

I agree with the "temporary workaround".

Will HP take a look and create a new version of package

Regards.

[Exploitation67]

+1

We have this problem too on our server.

Putting "nopat" on the kernel boot arguments helped.

I agree with the "temporary workaround".

Will HP take a look and create a new version of package

Regards.

Exploit

[rholighaus]

We have the same problem since upgrading our servers to Proxmox 5.0, which uses a modified Debian 9 (Stretch) kernel.

Using the nopat kernel option is a work-around but a fix (or repository) for Debian Stretch would be appreciated.

[roryt]

+1

We have the same problem on four new DL360 with Debian Stretch.

First of all the development has stoped in Jessie. The other utilities are working on stretch, but hp-health (which is vital for checking the status with scripts) seg-faults. Nopat is not a solution, the performance is severly degraded. HP please recompile for newer kernels !

[Jimmy Vance]

---

@roryt wrote:

+1

We have the same problem on four new DL360 with Debian Stretch.

First of all the development has stoped in Jessie. The other utilities are working on stretch, but hp-health (which is vital for checking the status with scripts) seg-faults. Nopat is not a solution, the performance is severly degraded. HP please recompile for newer kernels !

---

Keep in mind these forums are supported by the user community.  Some of us that work for HPE do monitor here when we can. 

To contact HPE directly on these Debian components you can use the information provided on the Linux Software Delivery Repository (HPE SDR) page.  The MCP covers Ubuntu, Debian, and other community supported distributions

https://downloads.linux.hpe.com/SDR/  The FAQ lists an email address to contact the group responsible Linux_SWdeliverables@groups.ext.hpe.com

With all that being said, here is the response you would receive

"The 10.60 MCP components are being compiled now, going into QA.   We compile on Ubuntu LTS,
then test on Debian (stretch in this case).  While we target 90 days from an Ubuntu LTS release.  I don't have an exact date for you at this time."

[norvikb]

What about CentOS7?

I have same crash on HPE DL360 G7 with CentOS 7

[Komputnik]

Hi,

half a year and no update ?

The 10.60 version also not fixes the problem on x86_64 centos 7.

Only nopat works.

When will it be fixed ?

Rgds.

Komputnik

[CMertes]

Is there any update on this after more than 2 month? For us the nopat option works for in generall, but brakes our GPUs on the server. Therefore the nopat workaround is no a real option for us. 

And a working healthcheck through monitoring software like nagios or icinga is important for us. We are running on x86_64 centos 7.

[Jimmy Vance]

---

wrote:

Is there any update on this after more than 2 month? For us the nopat option works for in generall, but brakes our GPUs on the server. Therefore the nopat workaround is no a real option for us. 

And a working healthcheck through monitoring software like nagios or icinga is important for us. We are running on x86_64 centos 7.

---

Fromwhat I've been told, this should have been resolved with v10.60.  I just loaded CentOS 7.3 on a DL20 Gen9 along with hp-health v10.60 and see no error.  I also did a yum update to get the latest packages and still had no error.  What kernel are you running?

[marzlarz]

We are seeing the same issue after upgrading to the latest kernel in CentOS 7.4  ( kernel-3.10.0-693.17.1.el7.x86_64 )