From: John Feeney <jfeeney@redhat.com> Date: Wed, 21 May 2008 11:13:28 -0400 Subject: [acpi] remove processor module errors Message-id: 48343C18.6030101@redhat.com O-Subject: [RHEL5.3 PATCH] rm acpi processor module errors Bugzilla: 228836 RH-Acked-by: Prarit Bhargava <prarit@redhat.com> RH-Acked-by: Brian Maly <bmaly@redhat.com> bz228836 https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=228836 acpi processor module displays errors if hyperthreading disabled Description of problem: If hyperthreading is disabled via bios (such as on a Dell Precision 490), an acpi_exception is hit on boot ("AE_NOT_FOUND, Processor Device is not present"). A relatively uninformative printk is also seen ("Getting cpuindex for acpiid"). Solution: acpi_processor_get_info() calls acpi_evaluate_object() to find the acpi_id for the processor. cpu_index is returned by convert_acpid_to_cpu(). Refer to drivers/acpi/processor_core.c. If hyperthreading is enabled, cpu_index has a non -1 value for each call and the code continues on without calling acpi_processor_hotadd_init(). When hyperthreading is disabled, cpu_index is set to -1 for the half the calls to acpi_processor_get_info(). As a result, the code calls acpi_processor_hotadd_init() when cpu_index is -1 and is_processor_present() is then called. is_processor_present() calls for status (_STA) to determine if the processor is present which it isn't in this situation and ACPI_EXCEPTION() is called. Three upstream patches have addressed this issue. The first changed the logic to look for a successful return and add some logic to better deal with failure. A second patch modified the first so an exception was not taken if the additional processor was not found. Since obtaining status is required for hotplugging, a debug printk was be used to indicate hotplugging is not an option if acpi_evaluate_integer() function failed, as opposed to an exception since there really is nothing wrong. The third patch removed the uninformative printk in acpi_processor_get_info(). Upstream status: These three patches are documented below: 1. Change check from FAILURE to SUCCESS cfaf3747ff3d431fba33f75083b7f50f58ae22ff ACPI Exception (acpi_processor-0677): AE_NOT_FOUND, Processor Device is not present According to the ACPI spec 6.3.7, "If a device object (including the processor object) does not have an _STA object, then OSPM assumes that all of the above bits are set,(in other words, the device is present, enabled, shown in the UI and funtioning)". is_processor_present shoud return 1 if the processor device object exists while it doesn't have an _STA object. http://bugzilla.kernel.org/show_bug.cgi?id=8570 Signed-off-by: Zhang Rui <rui.zhang@intel.com> Signed-off-by: Len Brown <len.brown@intel.com> 2. Change above so don't get the exception in all failure cases. d0ce46f550ebbd765881e8c48f43b66285d798b0 update cfaf3747ff3d431fba33f75083b7f50f58ae22ff <http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=object;h=cfaf3747ff3d431fba33f75083b7f50f58ae22ff> ACPI: ACPI Exception (): AE_NOT_FOUND, Processor Device is not present is_processor_present is only called in the processor hotplug case, and _STA method is mandatory at this time. We should ignore those processors that are disabled in the MADT and don't have _STA methods. Because they will never exist in this system. For the processors that don't physically exist but can be hot plugged later, we still need this debug info. http://bugzilla.kernel.org/show_bug.cgi?id=8570 Signed-off-by: Zhang Rui <rui.zhang@intel.com> Signed-off-by: Len Brown <len.brown@intel.com> 3. Remove the printk. d6637b28ffb38f207015c990e481fde5bba233d7 ACPI: delete two spurious ACPI messages <http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=d6637b28ffb38f207015c990e481fde5bba233d7> Venkatesh Pallipadi [Thu, 11 Jan 2007 04:16:36 +0000 (23:16 -0500)] ACPI: Getting cpuindex for acpiid 0x4 acpi_processor-0742 [00] processor_preregister_: Error while parsing _PSD domain information. Assuming no coordination http://bugzilla.kernel.org/show_bug.cgi?id=7286 Note: I combined the three upstream patches into one patch because upstream patch 1 and 2 dealt with the same code and third just removed two printks. Testing: The resulting patch was built with Brew (task 1325495). I tested the rpms successfully when hyperthreading was enabled and then disabled. I also successfully tested it on a system where hyperthreading could not be disabled. Acks would be appreciated. Thanks. diff --git a/drivers/acpi/processor_core.c b/drivers/acpi/processor_core.c index b13d644..6b206e2 100644 --- a/drivers/acpi/processor_core.c +++ b/drivers/acpi/processor_core.c @@ -476,9 +476,6 @@ static int acpi_processor_get_info(struct acpi_processor *pr) if (cpu_index == -1) { if (ACPI_FAILURE (acpi_processor_hotadd_init(pr->handle, &pr->id))) { - printk(KERN_ERR PREFIX - "Getting cpuindex for acpiid 0x%x\n", - pr->acpi_id); return -ENODEV; } } @@ -677,11 +674,19 @@ static int is_processor_present(acpi_handle handle) status = acpi_evaluate_integer(handle, "_STA", NULL, &sta); - if (ACPI_FAILURE(status) || !(sta & ACPI_STA_PRESENT)) { - ACPI_EXCEPTION((AE_INFO, status, "Processor Device is not present")); - return 0; - } - return 1; + + if (ACPI_SUCCESS(status) && (sta & ACPI_STA_DEVICE_PRESENT)) + return 1; + /* + * _STA is mandatory for a processor that supports hot plug + */ + if (status == AE_NOT_FOUND) + ACPI_DEBUG_PRINT((ACPI_DB_INFO, + "Processor does not support hot plug\n")); + else + ACPI_EXCEPTION((AE_INFO, status, + "Processor Device is not present")); + return 0; } static diff --git a/drivers/acpi/processor_perflib.c b/drivers/acpi/processor_perflib.c index 6fd174a..2f62eb2 100644 --- a/drivers/acpi/processor_perflib.c +++ b/drivers/acpi/processor_perflib.c @@ -736,10 +736,6 @@ int acpi_processor_preregister_performance( } err_ret: - if (retval) { - ACPI_DEBUG_PRINT((ACPI_DB_ERROR, "Error while parsing _PSD domain information. Assuming no coordination\n")); - } - for_each_possible_cpu(i) { pr = processors[i]; if (!pr || !pr->performance)