Hi all, I have a problem shutting down my computer: after selecting "shut down" from within KDE, or executing "shutdown -h" from the command line, the shutdown procedure reaches the point "system will be halted immediately...", but then it just hangs there and I have to turn off the computer by hand. Strangely, this does not happen all the time. Approx. a third of the time, shutdown works properly. Also, from Windows 2000 shutdown works without problems. Any hints, pointers, solutions etc? Here are my system specs: Intel Pentium4 2.4c processor (with HT) MSI Neo-2 LS mainboard running SuSE 9.0 with all patches installed (I had the same problem with SuSE 8.2, though) Kernel: vmlinuz-2.4.21-121-smp4G (had vmlinuz-2.4.21-99-smp4G installed previously, showing the same problem) Thanks Klaus __________________________________ Do you Yahoo!? Protect your identity with Yahoo! Mail AddressGuard http://antispam.yahoo.com/whatsnewfree
The Tuesday 2003-11-18 at 13:53 -0800, K V wrote:
procedure reaches the point "system will be halted immediately...", but then it just hangs there and I have to turn off the computer by hand. ... running SuSE 9.0 with all patches installed (I had the same problem with SuSE 8.2, though)
Yes, my 8.2 does it every time. Try to switch to console #10 during the halt sequence; in my case, I see a pair of acpi errors. I reported it to SuSE, they answered, but I see that they haven't solved it, apparently. -- Cheers, Carlos Robinson
On Tuesday 18 November 2003 18:42, Carlos E. R. wrote:
The Tuesday 2003-11-18 at 13:53 -0800, K V wrote:
procedure reaches the point "system will be halted immediately...", but then it just hangs there and I have to turn off the computer by hand.
...
running SuSE 9.0 with all patches installed (I had the same problem with SuSE 8.2, though)
Yes, my 8.2 does it every time. Try to switch to console #10 during the halt sequence; in my case, I see a pair of acpi errors. I reported it to SuSE, they answered, but I see that they haven't solved it, apparently.
Check to make sure that acpid is not unloading the ACPI modules as you shutdown, that can cause the problems you're seeing. When I first loaded 8.2, I modified the acpid settings in the run-level editor for just that reason. Mark Almeida -- Powered by SuSE Linux Pro 8.2/Kmail 1.5.4
The Tuesday 2003-11-18 at 19:35 -0800, The Wizard wrote:
Yes, my 8.2 does it every time. Try to switch to console #10 during the halt sequence; in my case, I see a pair of acpi errors. I reported it to SuSE, they answered, but I see that they haven't solved it, apparently.
Check to make sure that acpid is not unloading the ACPI modules as you shutdown, that can cause the problems you're seeing. When I first loaded 8.2, I modified the acpid settings in the run-level editor for just that reason.
I already tried. I have, in script '/etc/init.d/acpid' commented out the whole "stop" section. I may try to remove acpid altogether to see what happens. The error I get is: ACPI-1121: *** Error: Method execution failed [\SALD] (node efe3dc0), AE_AML_UNINITIALIZED_LOCAL ACPI-1121: *** Error: Method execution failed [\PTS] (node efed5c0), AE_AML_UNINITIALIZED_LOCAL I tracked that error message to a kernel source file, I don't remember which - not that I could see what the error is in plain language, anyway.
Mark Almeida --
Powered by SuSE Linux Pro 8.2/Kmail 1.5.4
-- Cheers, Carlos Robinson
On Tuesday 18 November 2003 22:53, K V wrote:
the shutdown procedure reaches the point "system will be halted immediately...", but then it just hangs there and I have to turn off the computer by hand.
Strangely, this does not happen all the time. Approx. a third of the time, shutdown works properly. Also, from Windows 2000 shutdown works without problems.
I have the same problem. It goes away if I use the k_deflt-2.4.21-99 kernel instead of the k_smp4G-2.4.21-99 kernel. This means that I loose hyperthreading, but the machine shuts down at least. I've also tried with the 121-kernel from mantel-kernels. Intel 865G Chipset, Abit mainboard, P4 2.6GHz -- Øystein Olsen, oystein.olsen@astro.uio.no, http://folk.uio.no/oeysteio Institute of Theoretical Astrophysics, http://www.astro.uio.no University of Oslo, Norway
On Wednesday 19 November 2003 03:00, Øystein Olsen wrote:
On Tuesday 18 November 2003 22:53, K V wrote:
the shutdown procedure reaches the point "system will be halted immediately...", but then it just hangs there and I have to turn off the computer by hand.
Strangely, this does not happen all the time. Approx. a third of the time, shutdown works properly. Also, from Windows 2000 shutdown works without problems.
I have the same problem. It goes away if I use the k_deflt-2.4.21-99 kernel instead of the k_smp4G-2.4.21-99 kernel. This means that I loose hyperthreading, but the machine shuts down at least. I've also tried with the 121-kernel from mantel-kernels.
Intel 865G Chipset, Abit mainboard, P4 2.6GHz
Have you guys looked into the settings in Yast for Modules NOT to unload? They have this covered. You can either do it in yast or surf to /etc/sysconfig/powermanagement and set this line in that file: ACPI_MODULES_NOT_TO_UNLOAD="ac battery button fan processor thermal" -- _____________________________________ John Andersen
On Thursday 20 November 2003 07:59, John Andersen wrote:
On Wednesday 19 November 2003 03:00, Øystein Olsen wrote:
On Tuesday 18 November 2003 22:53, K V wrote:
the shutdown procedure reaches the point "system will be halted immediately...", but then it just hangs there and I have to turn off the computer by hand.
Strangely, this does not happen all the time. Approx. a third of the time, shutdown works properly. Also, from Windows 2000 shutdown works without problems.
I have the same problem. It goes away if I use the k_deflt-2.4.21-99 kernel instead of the k_smp4G-2.4.21-99 kernel. This means that I loose hyperthreading, but the machine shuts down at least. I've also tried with the 121-kernel from mantel-kernels.
Intel 865G Chipset, Abit mainboard, P4 2.6GHz
Have you guys looked into the settings in Yast for Modules NOT to unload? They have this covered. You can either do it in yast or surf to /etc/sysconfig/powermanagement and set this line in that file: ACPI_MODULES_NOT_TO_UNLOAD="ac battery button fan processor thermal"
Tried it, but it doesn't work. Appending "acpismp=force apm=power-off" as a boot option to grub works, but then there is no acpi available at all. During boot, I get messages saying "no acpi in kernel". -- Øystein Olsen, oystein.olsen@astro.uio.no, http://folk.uio.no/oeysteio Institute of Theoretical Astrophysics, http://www.astro.uio.no University of Oslo, Norway
The Wednesday 2003-11-19 at 21:59 -0900, John Andersen wrote:
Have you guys looked into the settings in Yast for Modules NOT to unload? They have this covered. You can either do it in yast or surf to /etc/sysconfig/powermanagement and set this line in that file: ACPI_MODULES_NOT_TO_UNLOAD="ac battery button fan processor thermal"
Yes; but instead of doing it there, to be absolutely sure that I don't overlook any modules, I disabled the whole "stop" section in '/etc/init.d/acpid'. No change. Today, I have disabled the loading of modules in that script, and leaving only the 'startproc $ACPID_BIN' section. We'll see what happens. The error I'm getting is: ACPI-1121: *** Error: Method execution failed [\SALD] (node efe3dc0), AE_AML_UNINITIALIZED_LOCAL ACPI-1121: *** Error: Method execution failed [\PTS] (node efed5c0), AE_AML_UNINITIALIZED_LOCAL I'm tracking it to his file (grepping with mc): /usr/src/linux/drivers/acpi/dispatcher/dsmthdat.c And the code is this: --------------+++ acpi_status acpi_ds_method_data_get_value ( .... /* Examine the returned object, it must be valid. */ if (!object) { /* * Index points to uninitialized object. * This means that either 1) The expected argument was * not passed to the method, or 2) A local variable * was referenced by the method (via the ASL) * before it was initialized. Either case is an error. */ switch (opcode) { case AML_ARG_OP: ACPI_DEBUG_PRINT ((ACPI_DB_ERROR, "Uninitialized Arg[%d] at node %p\n", index, node)); return_ACPI_STATUS (AE_AML_UNINITIALIZED_ARG); case AML_LOCAL_OP: ACPI_DEBUG_PRINT ((ACPI_DB_ERROR, "Uninitialized Local[%d] at node %p\n", index, node)); ----> return_ACPI_STATUS (AE_AML_UNINITIALIZED_LOCAL); default: return_ACPI_STATUS (AE_AML_INTERNAL); } } ------------------- The error message is getting printed in drivers/acpi/parser/psparse.c or in drivers/acpi/utilities/uteval.c. But not by ACPI_DEBUG_PRINT... ah, that's a compile time switch (acpi section). I'm recompiling the kernel right now to check this out. Let's hope! :-) -- Cheers, Carlos
The Thursday 2003-11-20 at 14:12 +0100, Carlos E. R. wrote:
The error message is getting printed in drivers/acpi/parser/psparse.c or in drivers/acpi/utilities/uteval.c. But not by ACPI_DEBUG_PRINT... ah, that's a compile time switch (acpi section). I'm recompiling the kernel right now to check this out.
Let's hope! :-)
Well, the error message has changed a bit. It was:
ACPI-1121: *** Error: Method execution failed [\SALD] (node efe3dc0), AE_AML_UNINITIALIZED_LOCAL ACPI-1121: *** Error: Method execution failed [\PTS] (node efed5c0), AE_AML_UNINITIALIZED_LOCAL
Now it is (copied by hand, of course: computer hangs at that point): flushing ide drives... (HD claks softly about here) Power down. dsmthdat-0475 [29] ds_method_data_get_val: Uninitialized Local[0] at node eff61908 psparse-1121: *** Error: Method execution failed [\SALD] (node c193c728), AE_AML_UNINITIALIZED_LOCAL psparse-1121: *** Error: Method execution failed [\PTS] (node effec928), AE_AML_UNINITIALIZED_LOCAL And then it hangs (and keyboard locks). What I know is that it definitely is a programming bug, the debug message is too criptical, and that I can not correct it :-( -- Cheers, Carlos Robinson
Have you guys looked into the settings in Yast for Modules NOT to unload? They have this covered. You can either do it in yast or surf to /etc/sysconfig/powermanagement and set this line in that file: ACPI_MODULES_NOT_TO_UNLOAD="ac battery button fan processor thermal"
I checked, and the settings are just as you quoted. However, while browsing the SuSE support database I noticed they recommend including the kernel option "apm=real-mode-poweroff" or simply "apm=power-off" for multi-processor machines (which the ht-enabled systems are in some way). I set the kernel option "apm=real-mode-poweroff", and for the last two days shutdown worked. Might be just the luck of the draw, of course, since it used to work about 30% of the time. I'll keep the list updated... Klaus __________________________________ Do you Yahoo!? Free Pop-Up Blocker - Get it now http://companion.yahoo.com/
The Thursday 2003-11-20 at 14:06 -0800, K V wrote:
However, while browsing the SuSE support database I noticed they recommend including the kernel option "apm=real-mode-poweroff" or simply "apm=power-off" for multi-processor machines (which the ht-enabled systems are in some way).
But that dissables acpi for anything else, if I remember correctly. -- Cheers, Carlos Robinson
-----Original Message-----
From: "Carlos E. R."
The Thursday 2003-11-20 at 14:06 -0800, K V wrote:
However, while browsing the SuSE support database I noticed they recommend including the kernel option "apm=real-mode-poweroff" or simply "apm=power-off" for multi-processor machines (which the ht-enabled systems are in some way).
But that dissables acpi for anything else, if I remember correctly.
No, acpi is not the same as apm. Ken Schneider
The Friday 2003-11-21 at 06:18 -0500, Ken Schneider wrote:
The Thursday 2003-11-20 at 14:06 -0800, K V wrote:
However, while browsing the SuSE support database I noticed they recommend including the kernel option "apm=real-mode-poweroff" or simply "apm=power-off" for multi-processor machines (which the ht-enabled systems are in some way).
But that dissables acpi for anything else, if I remember correctly.
No, acpi is not the same as apm.
Right, and also they are incompatible: enabling one forces disabling the other. Thus, the above switch must enable apm, and if that is so, acpi is disabled. -- Cheers, Carlos Robinson
I set the kernel option "apm=real-mode-poweroff", and for the last two days shutdown worked. Might be just the luck of the draw, of course, since it used to work about 30% of the time. I'll keep the list updated...
Argh! I spoke too early. Shutdown didn't work today. Now I installed the default kernel, and I think I'm just gonna stick with that until a better solution comes along...
Klaus
__________________________________ Do you Yahoo!? Free Pop-Up Blocker - Get it now http://companion.yahoo.com/
participants (7)
-
Carlos E. R. -
John Andersen -
K V -
Ken Schneider -
pinto -
The Wizard -
Øystein Olsen