From Bugzilla Helper: User-Agent: Mozilla/5.0 (compatible; Konqueror/3.0.0; Linux) Description of problem: Rebooting Skipjack causes the Sony Vaio PCG-FXA36 laptop to be stuck at the BIOS screen with the IDE led stuck on. It sometimes shows a keyboard stuck error, but most of the time shows a graphical Sony logo that completely blocks obscures the BIOS text info. Testing: kernel-2.4.7-10.*.rpm FAILS to boot with kernel oopses. kernel-2.4.9-31.athlon.rpm HANGS on reboot kernel-2.4.18-0.[4-12].athlon.rpm HANGS on reboot kernel-2.4.9-31.i686.rpm SUCCESS in reboot kernel-2.4.18-0.[4-12].i686.rpm SUCCESS in reboot It seems that Red Hat i686 kernel packages do not exhibit this problem, while athlon kernel packages do. It may be notable to mention that both i586 and athlon arch packages of Mandrake 8.2 do not reboot properly, so they may be compiling in different options that trigger this. Version-Release number of selected component (if applicable): skipjack-beta1 How reproducible: Always Additional info: 1. Mandrake 8.2 on this machine has the same problem. 2. MandrakeForum has some reports of non-laptop Athlon machines exhibiting this behavior. 3. eric owns the same model Sony FXA-36 laptop and has reported the same problem to the Skipjack mailing list. (All current FXA series laptops have the same motherboard chipset, but different processors.)
I forgot to mention that "noathlon" kernel option did not change test results of the failed kernels above.
Created attachment 51324 [details] "dmesg" output
Created attachment 51325 [details] "lspci -v" output
Created attachment 51326 [details] dmidecode output (suggested by Mandrake kernel developer)
This problem did not exist in earlier athlon arch kernels. Would it be possible for me to have access to older athlon Rawhide kernel builds in order to do testing to find when this bug was introduced?
This page http://www.myplc.com/sony/vaio_unix_guide.htm thinks that this is a hard drive controller issue. One guy recompiled the Rawhide kernel with experimental ACPI and it prevented the reboot freeze. Installing the i686 arch kernel rather than athlon arch also prevents the reboot freeze.
Still exists in Red Hat 7.3 release.
I have the same problem on 7.2 and 7.3
For now you can use the i686 kernel to avoid this reboot bug. Alternatively you can compile yourself an ACPI capable kernel, although you will want to use the latest 2.4.19-pre- Alan Cox kernel to have a working ACPI implementation that doesn't crash.
I fail to see how ACPI could make *ANY* difference in this. The only difference between the athlon and i686 kernel is how memcpy() is implemented....
APM athlon gets stuck in reboot with IDE light stuck on. APM i686 reboots fine. ACPI athlon reboots fine. ACPI i686 reboots fine. I can't explain it. I only know this through empirical testing.
Compaq Presario 700 series with Mobile Athlon or Duron processor has the same hang in BIOS with HD light stuck on after rebooting from Red Hat athlon optimized kernel. rpm --force installing the i686 kernel RPM fixes this problem on this series of laptops too. This may be more generally a "Mobile Athlon w/ certain VIA Chipset" problem.
Confirmed still hangs after reboot with athlon kernel-2.4.18-7.80, but not i686 kernel-2.4.18-7.80 of Limbo2.
Ok I think I have a handle on this, it seems you bios doesn't like the LOCAL APIC config option, which happened to be turned on for Athlon and off for i686
Would a permanent solution in the kernel be possible, or we must always use a the i686 workaround?
I've disabled local apic for athlon kernels too in the config, that'll be in the next build
kernel-2.4.18-0.75.athlon reboot SUCCESS kernel-2.4.18-0.75.i686 reboot FAIL kernel-2.4.18-0.75.i586 reboot SUCCESS kernel-2.4.18-0.75.i386 reboot SUCCESS Should close bug?
Anaconda using kernel-2.4.18-10.99 locks up after reboot. Can this fix be applied to that kernel too?
So, looks like we still need to do something with the boot kernel, but since we don't run separate boot kernels (per arch) not sure what the correct fix is here. Reopening to get an answer.
Unfortionatly we can't do this for the BOOT kernel since that would disable the workaround needed for buggy 440GX chipset bioses ;(
Reproduced this problem on the Compaq Presario 900US. This laptop has an ALi chipset, unlike the Sony Vaio FXA series and Presario 700 series which both have VIA. This may be indicative that this is more generally a Mobile Athlon kernel problem. Unrelated bug: This Presario 900 series laptop seems completely inoperative with the ALI IDE kernel drivers, failing after over a minute of attempting to access the hard drive or CD-ROM IDE devices. I am testing it with various other kernels tomorrow.
If the ALi problem is present in the Null beta kernel please open a seperate bug for it and stick alan in the cc line. Thanks
Bug 72387: (Compaq Presario 900) ALI15X3 IDE fails Bug 72388: (Compaq Presario 900) 8139too module fails
Hi: I have a Sony Vaio FXA47 with Red Hat 8.0 and kernel 2.4.18-17.8.0. Guess what, systems still hangs when rebooting, displaying the Sony logo and the LED is stuck. IF you wait some time, the BIOS eventually tells you that there was a keyboard error. I checked the config and it says "Local APIC support on uniprocessors" is disabled, but the problem persists. Any chances on seeing this issue fixed? Thanks, Luis Delgado (Curly Howard) pangair
pangair, Are you sure that you installed the athlon arch kernel?
How do I check that out? Thanks.
pangair, Sorry I'm not exactly sure how to check which arch kernel is currently running. Arjan will know. Oh BTW, could you please read Bug 74315. Our laptops are virtually the same except for the BIOS revision (which you can flash to be the same.) Could you please read Bug 74315 and check if you experience this problem too? Thanks, Warren
Within Phoebe, still locks up in BIOS after rebooting from the BOOT kernel, but athlon arch kernel is fine.
supposed to be fixed in recent kernels; apic is off for the BOOT kernel
Phoebe3's installer no longer locked up after installation. Seems fixed.