Messages in this thread Patch in this message |  | | Date | Mon, 24 Sep 2001 17:29:10 +0200 | From | Pavel Machek <> | Subject | Workaround for ACPI sleep problems |
| |
Hi!
Some machines have bad problems with acpi... Toshiba will kill interrupts and mvp3 gives totally incorrect table, meaning that sleeps just will not work.
I figured part of table, but I *know* there's working S1 somewhere out there, so if someone has the correct value to use, please let me know. [If you replace number near /* HERE */ with correct value in 0-15 range, you'll get working S1. It's only 15 tries ;-).]
This fixes acpi on toshiba, and provides at least working poweroff on MVP3. Pavel
--- clean/include/asm-i386/system.h Sun Sep 23 23:07:00 2001 +++ linux/include/asm-i386/system.h Mon Sep 24 16:39:54 2001 @@ -340,4 +340,8 @@ void disable_hlt(void); void enable_hlt(void); +extern unsigned long dmi_broken; +#define BROKEN_ACPI_Sx 0x0001 +#define BROKEN_INIT_AFTER_S1 0x0002 + #endif --- clean/arch/i386/kernel/dmi_scan.c Sun Sep 23 23:04:49 2001 +++ linux/arch/i386/kernel/dmi_scan.c Mon Sep 24 17:01:00 2001 @@ -10,6 +10,8 @@ #include <linux/keyboard.h> #include <asm/keyboard.h> +unsigned long dmi_broken; + struct dmi_header { u8 type; @@ -365,6 +367,38 @@ } /* + * ASUS K7V-RM has broken ACPI table defining sleep modes + */ + +static __init int broken_acpi_Sx(struct dmi_blacklist *d) +{ + printk(KERN_WARNING "Detected ASUS mainboard with broken ACPI sleep table\n"); + dmi_broken |= BROKEN_ACPI_Sx; + return 0; +} + +/* + * Toshiba keyboard likes to repeat keys when they are not repeated. + */ + +static __init int broken_toshiba_keyboard(struct dmi_blacklist *d) +{ + printk(KERN_WARNING "Toshiba with broken keyboard detected. If your keyboard sometimes generates 3 keypresses instead of one, contact pavel@ucw.cz\n"); + return 0; +} + +/* + * Toshiba fails to preserve interrupts over S1 + */ + +static __init int init_ints_after_s1(struct dmi_blacklist *d) +{ + printk(KERN_WARNING "Toshiba with broken S1 detected.\n"); + dmi_broken |= BROKEN_INIT_AFTER_S1; + return 0; +} + +/* * Some Bioses enable the PS/2 mouse (touchpad) at resume, even if it * was disabled before the suspend. Linux gets terribly confused by that. */ @@ -543,7 +577,19 @@ MATCH(DMI_PRODUCT_NAME, "Dell PowerEdge 8450"), NO_MATCH, NO_MATCH, NO_MATCH } }, - + { broken_acpi_Sx, "ASUS K7V-RM", { /* Bad ACPI Sx table */ + MATCH(DMI_BIOS_VERSION,"ASUS K7V-RM ACPI BIOS Revision 1003A"), + MATCH(DMI_BOARD_NAME, "<K7V-RM>"), + NO_MATCH, NO_MATCH + } }, + { broken_toshiba_keyboard, "Toshiba Satellite 4030cdt", { /* Keyboard generates spurious repeats */ + MATCH(DMI_PRODUCT_NAME, "S4030CDT/4.3"), + NO_MATCH, NO_MATCH, NO_MATCH + } }, + { init_ints_after_s1, "Toshiba Satellite 4030cdt", { /* Reinitialization of 8259 is needed after S1 resume */ + MATCH(DMI_PRODUCT_NAME, "S4030CDT/4.3"), + NO_MATCH, NO_MATCH, NO_MATCH + } }, { NULL, } }; --- clean/drivers/acpi/hardware/hwregs.c Sun Sep 23 23:05:16 2001 +++ linux/drivers/acpi/hardware/hwregs.c Mon Sep 24 17:15:58 2001 @@ -29,6 +29,7 @@ #include "acpi.h" #include "achware.h" #include "acnamesp.h" +#include <asm/system.h> #define _COMPONENT ACPI_HARDWARE MODULE_NAME ("hwregs") @@ -162,6 +163,22 @@ if ((sleep_state > ACPI_S_STATES_MAX) || !slp_typ_a || !slp_typ_b) { return_ACPI_STATUS (AE_BAD_PARAMETER); + } + + if (dmi_broken & BROKEN_ACPI_Sx) { + printk("Attempted to enter %d\n", sleep_state); + switch (sleep_state) { + /* 0,3,4,6,7,11 is nop */ + /* 5 is nop, too? No. 5 is some valid state, just not S1 nor S3 (strange). */ + /* 13 is lockup when used in S1 or S3 context */ + /* 14 is "keep monitor on and than all goes off when power is pressed in S1 context */ + case 1: *slp_typ_a = *slp_typ_b =12 /* HERE */; return_ACPI_STATUS (status); + case 3: *slp_typ_a = *slp_typ_b =14; return_ACPI_STATUS (status); + case 4: *slp_typ_a = *slp_typ_b = 1; return_ACPI_STATUS (status); /* Seems ok, but it times somehow long to wakeup. Is it possible this is S3? */ + case 5: *slp_typ_a = *slp_typ_b =10; return_ACPI_STATUS (status); /* Seems ok */ + default: return_ACPI_STATUS (AE_BAD_PARAMETER) + } + } /* --- clean/drivers/acpi/ospm/system/sm_osl.c Sun Sep 23 23:05:21 2001 +++ linux/drivers/acpi/ospm/system/sm_osl.c Mon Sep 24 16:41:14 2001 @@ -719,6 +780,11 @@ /* TODO: resume devices and restore their state */ enable(); + + if (dmi_broken & BROKEN_INIT_AFTER_S1) { + printk("Broken toshiba laptop -> kicking interrupts\n"); + init_8259A(0); + } return status; } -- I'm pavel@ucw.cz. "In my country we have almost anarchy and I don't care." Panos Katsaloulis describing me w.r.t. patents at discuss@linmodems.org - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
|  |