Menu
▾
▴
Re: [ipmiutil-developers] Win64: heap leak crashing ipmiutil 3.13
|
From: Andy C. <and...@gm...> - 2019-07-16 21:48:44
|
If this is SuperMicro firmware it could be another firmware failure. In fact, the previous symptom with ipmiutil fru was due to a SuperMicro firmware bug, causing the firmware to fail on SuperMicro X11DPT-B. I just realized that I didn't reply all, so I copied that email below. A firmware crash can happen on SuperMicro firmware (e.g. threshold exceeded and it cannot reconcile the fluctuations, or a few other bugs). If that happens, warm booting the firmware may be possible via 'ipmiuilt reset -k', but otherwise you will need to remove input power for 10 seconds to power up the firmware cleanly. Andy ---------- Forwarded message --------- From: Andy Cress <and...@gm...> Date: Thu, Jul 11, 2019 at 9:16 PM Subject: Re: [ipmiutil-developers] Win64: heap leak crashing ipmiutil 3.13 To: Rob Scheepens <rob...@nu...> I see the problem. The firmware drops the session with a partially filled buffer when doing the get_system_info command. There are a few firmware vendors/versions (e.g. Sun) where this is not supported, and I need to add this one to set do_systeminfo = 0 and skip that function. devid: firmware ver 6.49, IPMI v02, vendor=10876 prod=2402. (= SuperMicro X11DPT-B) It should be supported in all IPMI v2.0, but apparently is not working in this case. I had tested this on a bunch of different SuperMicro motherboards. Sigh. SuperMicro isn't always consistent. Technically this is a firmware bug, but it could be several years before they fix it. Are there any other SuperMicro systems that see this problem? Andy On Mon, Jul 15, 2019 at 8:32 PM Abhijit Sunil Betigeri < abh...@nu...> wrote: > Have encountered issue where ipmiutil,.exe hangs and none of commands > work, this includes ipmicfg-win as well. > Attaching the system event log. > > > ipmiutil fru -x > > ipmiutil fru version 3.11 > > ipmi_open: driver type = > > ipmi_open_ia: imbdrv request error, ret=1 ccode=c0 > > ipmi_open_ms: ObjectPath: Microsoft_IPMI.InstanceName="ACPI\\IPI0001\\0_0" > > ipmi_open rc = 0 type = ms > > Driver type ms, open rc = 0 > > ipmi_cmdraw_ms(cmd=1,netfn=6,lun=0,sa=20,sdata=0) RequestResponse ret=0 > > ipmi_cmdraw_ms: CompletionCode ff returned > > ipmi_cmdraw_ms: resp data(2): 00 00 > > ccode ff: Unspecified error > > ipmiutil fru, Unspecified error > > This implies to ipmi driver issue? isnt it? > > Regards, > > Abhijit > ------------------------------ > *From:* Rob Scheepens > *Sent:* Tuesday, July 9, 2019 11:35 PM > *To:* Andy Cress > *Cc:* ipm...@li...; Abhijit Sunil Betigeri; > Anupam Chakraborty; Naga Chandana > *Subject:* Re: [ipmiutil-developers] Win64: heap leak crashing ipmiutil > 3.13 > > > Using “ipmiutil fru -x” I get a 100% reproduction rate. The debug log of > one attempt is attached. All five attempts I did are the same: they stop at: > > ... > > get_sysinfo(1,2) j=2 len=15 > > ipmi_cmdraw_ms(cmd=59,netfn=6,lun=0,sa=20,sdata=4) RequestResponse ret=0 > > ipmi_cmdraw_ms: req data(4): 00 01 03 00 > > ipmi_cmdraw_ms: CompletionCode 80 returned > > ipmi_cmdraw_ms: resp data(2): 80 d0 > > ccode 80: Invalid Session Handle or Empty Buffer > > > > \Rob > > > > *From: *Andy Cress <and...@gm...> > *Date: *Wednesday, 10 July 2019 at 03:53 > *To: *Rob Scheepens <rob...@nu...> > *Cc: *"ipm...@li..." < > ipm...@li...>, Abhijit Sunil Betigeri < > abh...@nu...>, Anupam Chakraborty < > anu...@nu...>, Naga Chandana <nag...@nu...> > *Subject: *Re: [ipmiutil-developers] Win64: heap leak crashing ipmiutil > 3.13 > > > > > > Sure the source is available. Here is the link for that version. > > http://sourceforge.net/projects/ipmiutil/files/ipmiutil-3.1.3.tar.gz > [sourceforge.net] > <https://urldefense.proofpoint.com/v2/url?u=http-3A__sourceforge.net_projects_ipmiutil_files_ipmiutil-2D3.1.3.tar.gz&d=DwMFaQ&c=s883GpUCOChKOHiocYtGcg&r=OMged-t_5I_fmfpUaT3vaA06lgLL_alYnDQJxHmXz64&m=DIsjsjlFIOJRbicayrSFqCCQVzTlqRjaQ04bd8WjicM&s=46h52Eb8jp-lu4UCCD6qH7HnvJnVOWctoQ20aQcztPQ&e=> > > > > > One other clue would be to run this command on the system where the dumps > occur: > > ipmiutil fru -x > > and send me the (debug) output. > > The output would show a good bit of where it fails, if it fails > frequently. > > > > Andy > > > > On Mon, Jul 8, 2019 at 7:04 AM Rob Scheepens <rob...@nu...> > wrote: > > Hi Andy, > > > > The commandline is “'"C:\Program Files\sourceforge\ipmiutil\ipmiutil.exe" > fru' “. Reproduction is fairly reliable, see timestamps of the user dumps: > > > > 07/05/2019 12:38 AM 24,044,684 ipmiutil.exe-dumps.zip > > 07/08/2019 01:55 AM 65,382,554 ipmiutil.exe.10568.dmp > > 07/07/2019 11:55 PM 65,374,160 ipmiutil.exe.11612.dmp > > 07/07/2019 10:56 PM 65,370,628 ipmiutil.exe.13348.dmp > > 07/08/2019 02:56 AM 65,386,532 ipmiutil.exe.14328.dmp > > 07/08/2019 12:55 AM 65,378,874 ipmiutil.exe.15160.dmp > > 07/08/2019 01:57 AM 65,408,636 ipmiutil.exe.9272.dmp > > > > Instead of PDBs, can I get the source code somewhere and line it up in > WinDbg? I’ve uploaded two dumps to https://we.tl/t-FGq9I8scLA [we.tl] > <https://urldefense.proofpoint.com/v2/url?u=https-3A__we.tl_t-2DFGq9I8scLA&d=DwMFaQ&c=s883GpUCOChKOHiocYtGcg&r=OMged-t_5I_fmfpUaT3vaA06lgLL_alYnDQJxHmXz64&m=DIsjsjlFIOJRbicayrSFqCCQVzTlqRjaQ04bd8WjicM&s=n89zN5bGfGYqXi2WHJ2X2TXmMQ2QiJashFzYCJx8lSs&e=>. > > > > > Iirc we recently switched from imbdrv to ipmidrv because of an issue. > @Abhijit: can you (dis)confirm? > > > > \Rob > > |
