Most recent kernel where this bug did not occur: ? Distribution: openSUSE 10.2 Hardware Environment: Supermicro motherboard H8QC8+, Quad AMD Opteron 852 (x86_64) Software Environment: kernel.org 2.6.21.5 Problem Description: Kernel seems to have trouble registering and PCI-E Infiniband card (HP NC571C, Dual Port, 128 MB, PCI-E card). I've not yet used the card, so I cannot say if a more serious error is triggered - I was hoping to take one step at a time. Steps to reproduce: boot.msg shows 'kernel bug?' on boot-up. System Summary (more detials attached): * CPU: AMD, Opteron 852, 1.6GHz * Motherboard: Supermicro, H8QC8+ * BIOS: R1.0a, Ver: 8.00.12, Date: 04/10/2007, ID: 1HQC8008 * Chipset: nVidia nForce Pro 2200 (CK804) (Media and communications processor) * Chipset: AMD 8132 (HyperTransport PCI-X Tunnel) * RAM: 8 x 1GB ATP, PC3200 ECC REG Kernel 2.6.21.5 (www.kernel.org), the .config file is attached - essentially I use the openSUSE 10.2 config file with the following changes: General setup:Local version:append to kernel release:-opteron Processor type and features -Choose CPU type:AMD/opteron -Preemption model: No Forced preemption -Timer frequency: 1000Hz I ran the firmware test routines that are distributed on the openSUSE 10.2 DVD (their kernel 2.6.). My report of the failed tests and Supermicro's response are below. (I wasn't able to save the test results) > A) Running the linux-ready firmware tests (distributed with openSUSE > 10.2) shows that several tests fail: > 1) [Fail] DMI information check: > Reports: two bab index values and one out of spec value found > (no actual values are reported....) > > 2) [Fail] DSDT AML verification check > Reports: "String must be entirely alphanumeric (_NCRAIDBUS) > in PCI0.NVRB > There are 7 warnings: "Possible operator timeout is ignored > in \_SB.RBPE." > > 3) [Fail] General ACPI information > Reports twice: "PCI Express port driver reports an invalid > IRQ" > > 4) [Fail] (experimental) APIC Edge/Level check > Reports: "Non-legacy interrupt 0 incorrectly edge triggered > Non-legacy interrupt 1 incorrectly edge > triggered > Non-legacy interrupt 8 incorrectly edge > triggered > Non-legacy interrupt 14 incorrectly edge > triggered" > > > 5) [Fail] HPET configuration test > Reports: "Failed to locate HPET base" > > There are several warnings about: > - EDD Boot disk hinting > - ACPI passive thermal trip points > Reports: "Zone * doesn't support passive trip point at all" > - MCFG PCI Express* memory mapped config space: > Reports: "No MCFG ACPI table found. This table is required for > PCI Express*." > > Passed tests: > - MTRR validation > - FADT test > - Validate assigned PCI resources > - PCI Express MaxReadReq tuning > - OS/2 memory hole test > - Battery tests (No battery information present) > The Supermicro response was: <quote> Allen Gao has provided the attached BIOS for the SuSE firmware test issue. With the BIOS settings set to default, the following will be reported as FAILED: 1.) MTRR Validation 2.) APIC Edge/Level Check 3.) HPET Configuration Test The following is our response to these failures: 1.) With MTRR set to continuous in the BIOS, the test will display a failure. With MTRR set to discrete, this test will pass. The customer can change the default the setting to discrete in BIOS. 2.) According to our BIOS engineer, the interrupts are legacy devices and they should be edge triggered. The test says that they are incorrect as being edge triggered. 3.) The nVidia CK804 chipset does not support HPET. </quote> The mcelog is empty. I've extracted the attached the messages log, and some information obtained by running: frontend:~ # lspci &>/tmp/sm_h8qc8_lspci.txt frontend:~ # lspci -vvxx &>>/tmp/sm_h8qc8_lspci.txt frontend:~ # lsmod &>/tmp/sm_h8qc8_lsmod.txt frontend:~ # cat /proc/interrupts &>/tmp/sm_h8qc8_interrupts.txt frontend:~ # hwinfo &>/tmp/sm_h8qc8_hwinfo.txt frontend:~ # pnpdump &>/tmp/sm_h8qc8_pnpdump.txt frontend:~ # procinfo &>/tmp/sm_h8qc8_procinfo.txt
Created attachment 11950 [details] kernel build config file
Created attachment 11951 [details] Supermicro H8QC8 (2.6.21.5) boot.msg
Created attachment 11952 [details] Supermicro H8QC8 (2.6.21.5) hwinfo
Created attachment 11953 [details] Supermicro H8QC8 (2.6.21.5) interrupts
Created attachment 11954 [details] Supermicro H8QC8 (2.6.21.5) lsmod
Created attachment 11955 [details] Supermicro H8QC8 (2.6.21.5) lspci
Created attachment 11956 [details] Supermicro H8QC8 (2.6.21.5) messages
Created attachment 11957 [details] Supermicro H8QC8 (2.6.21.5) pnpdump
Created attachment 11958 [details] Supermicro H8QC8 (2.6.21.5) procinfo
BTW, since this is my first report, I should have noted my thanks for all the great work you guys have put into making such great software - I'm a 'user' rather than programmer and have just switched from WinXP... Linux continues to surprise and astonish me :) While my report relates to the 2.61.5 from kernel.org, the following report seems to show similar symptoms for Linux ubuntu 2.6.20-8-generic. https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.20/+bug/90863 Hope this helps?
I'm pretty sure the error reported above is related to the presense of the pci-e card. While I'm able to load the openfabrics software, and get the opensm running on each machine, the interfaces ib0 and ib1 come up without any problem reported, while trying to get IPoIB working I observed the following: frontend:~ # arping -I ib0 -c 3 192.168.0.32 *** buffer overflow detected ***: arping terminated ======= Backtrace: ========= /lib64/libc.so.6(__chk_fail+0x2f)[0x2aea79dfcd4f] arping(main+0x48a)[0x55555555692a] /lib64/libc.so.6(__libc_start_main+0xf4)[0x2aea79d47ae4] arping[0x555555555a89] ======= Memory map: ======== 2aea798f7000-2aea79913000 r-xp 00000000 fd:03 41944073 /lib64/ld-2.5.so 2aea79913000-2aea79914000 rw-p 2aea79913000 00:00 0 2aea79946000-2aea79947000 rw-p 2aea79946000 00:00 0 2aea79b13000-2aea79b15000 rw-p 0001c000 fd:03 41944073 /lib64/ld-2.5.so 2aea79b15000-2aea79b26000 r-xp 00000000 fd:03 41944108 /lib64/libresolv-2.5.so 2aea79b26000-2aea79d26000 ---p 00011000 fd:03 41944108 /lib64/libresolv-2.5.so 2aea79d26000-2aea79d28000 rw-p 00011000 fd:03 41944108 /lib64/libresolv-2.5.so 2aea79d28000-2aea79d2a000 rw-p 2aea79d28000 00:00 0 2aea79d2a000-2aea79e63000 r-xp 00000000 fd:03 41944080 /lib64/libc-2.5.so 2aea79e63000-2aea7a062000 ---p 00139000 fd:03 41944080 /lib64/libc-2.5.so 2aea7a062000-2aea7a065000 r--p 00138000 fd:03 41944080 /lib64/libc-2.5.so 2aea7a065000-2aea7a067000 rw-p 0013b000 fd:03 41944080 /lib64/libc-2.5.so 2aea7a067000-2aea7a06e000 rw-p 2aea7a067000 00:00 0 2aea7a06e000-2aea7a07b000 r-xp 00000000 fd:03 41977950 /lib64/libgcc_s.so.1 2aea7a07b000-2aea7a27a000 ---p 0000d000 fd:03 41977950 /lib64/libgcc_s.so.1 2aea7a27a000-2aea7a27c000 rw-p 0000c000 fd:03 41977950 /lib64/libgcc_s.so.1 555555554000-555555558000 r-xp 00000000 fd:03 92318767 /sbin/arping 555555757000-555555759000 rw-p 00003000 fd:03 92318767 /sbin/arping 555555759000-55555577a000 rw-p 555555759000 00:00 0 [heap] 7fff3119e000-7fff311b3000 rw-p 7fff3119e000 00:00 0 [stack] ffffffffff600000-ffffffffff601000 r-xp 00000000 00:00 0 [vdso] Aborted
I have downloaded and run the linux-firmware kit release 2. This time the usb save seems to have worked and I attach the files that were produced. the tests run were the default tests, i.e. those run after no keyboard response. The acpi.dump file was empty.
Created attachment 11969 [details] linux-firmware kit (release 2) DSDT.aml
Created attachment 11970 [details] linux-firmware kit (release 2) DSDT.dat
Created attachment 11971 [details] linux-firmware kit (release 2) DSDT.dat
>Unable to register resource 0x00000000ffc00000-0x00000000ffffffff - kernel >bug? Which file contains the error message? I can't find it. Can you tell me what failure you are seeing? the firmware kit test result sometimes isn't quilte useful.
Also can you attach the output of 'cat /proc/iomem'?
Hi David, Thanks for taking the time to look at this. The error msg appears in /var/log/boot.msg, Apologies for not attaching this - I thought I had. I will now also attach the output of `cat /proc/iomem`
Created attachment 12143 [details] Supermicro H8QC8 (2.6.21.5) proc iomem Information requested in comment #17
Re comment #16 and #18, I now see I did upload the boot.msg file. The error message is in the second file that was originally uploaded (around line 500-550?): Supermicro H8QC8 (2.6.21.5) boot.msg
For comment #18 the error I'm seeing so far is that in the boot.msg and the linux firmware kit report that there is an invalid interrupt request. Specifically, that described in the OP: > 3) [Fail] General ACPI information > Reports twice: "PCI Express port driver reports an invalid > IRQ" This [Fail] occurs with the LFK distributed on the openSUSE 10.2 CD's and the LFK release 2. More seriously I was not able run the openSUSE install - it hung at some time sorry I didn't record the details - this is going back sometime now. Supermicros' suggestion was to remove the card, install Linux then insert the card. I installed linux, upgraded the kernek and inserted the card - and here we are :) I haven't be courageous enough to try too much with the card, so can't report any more serious events than the one in comment #11 - which might be caused by something else....? Anyway, I was encouraged that I'm not alone in seeing this error - the following ubuntu reports might be helpful?: http://ubuntuforums.org/showthread.php?t=404751 https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.20/+bug/90863
>ck804xrom ck804xrom_init_one(): Unable to register resource >0x00000000ffc00000-0x00000000ffffffff - kernel bug? This doesn't break anything. It's just a warning, we could remove it completely. Did you see real breakage? >I haven't be courageous enough to try too much with the card, so can't report >any more serious events than the one in comment #11 - which might be caused by >something else....? It sounds application is broken, do you kernel breakage.
OK, thanks for the info. I reported this because it was flagged as a possible bug rather than just a warning, I haven't used the card yet for the same reason :) So, no there is no other breakage to report. You can close this bug then. Thanks again for the excellent work you guys do!
The message is really missleading, but it's not a bug. I'm closing the bug.