Re: BUG: linux 2.6.19 unable to enable acpi

[Len Brown <lenb@kernel.org>]

The code that enables ACPI mode hasn't really changed since before 2.6.12 -- 
unless udelay() has changed beneath us...
So if you are going to test an old version of Linux, you should start before then.

Perhaps you can try this debug patch on top of 2.6.19 and send along the dmesg?
(also, please include CONFIG_ACPI_DEBUG=y)

thanks,
-Len



diff --git a/drivers/acpi/hardware/hwacpi.c b/drivers/acpi/hardware/hwacpi.c
index de50fab..c782da3 100644
--- a/drivers/acpi/hardware/hwacpi.c
+++ b/drivers/acpi/hardware/hwacpi.c
@@ -119,6 +119,9 @@ acpi_status acpi_hw_set_mode(u32 mode)
 	 * we make sure both the numbers are zero to determine these
 	 * transitions are not supported.
 	 */
+printk("ACPI: FADT.acpi_enable %d\n", acpi_gbl_FADT->acpi_enable);
+printk("ACPI: FADT.acpi_disable %d\n", acpi_gbl_FADT->acpi_disable);
+
 	if (!acpi_gbl_FADT->acpi_enable && !acpi_gbl_FADT->acpi_disable) {
 		ACPI_ERROR((AE_INFO,
 			    "No ACPI mode transition supported in this system (enable/disable both zero)"));
@@ -130,6 +133,9 @@ acpi_status acpi_hw_set_mode(u32 mode)
 
 		/* BIOS should have disabled ALL fixed and GP events */
 
+printk("ACPI: smi_cmd 0x%x, acpi_enable 0x%x\n",
+	acpi_gbl_FADT->smi_cmd, (u32) acpi_gbl_FADT->acpi_enable);
+
 		status = acpi_os_write_port(acpi_gbl_FADT->smi_cmd,
 					    (u32) acpi_gbl_FADT->acpi_enable,
 					    8);
@@ -164,7 +170,7 @@ acpi_status acpi_hw_set_mode(u32 mode)
 	 * Some hardware takes a LONG time to switch modes. Give them 3 sec to
 	 * do so, but allow faster systems to proceed more quickly.
 	 */
-	retry = 3000;
+	retry = 3000 * 100;
 	while (retry) {
 		if (acpi_hw_get_mode() == mode) {
 			ACPI_DEBUG_PRINT((ACPI_DB_INFO,
@@ -175,6 +181,7 @@ acpi_status acpi_hw_set_mode(u32 mode)
 		acpi_os_stall(1000);
 		retry--;
 	}
+printk("ACPI: retry %d\n");
 
 	ACPI_ERROR((AE_INFO, "Hardware did not change modes"));
 	return_ACPI_STATUS(AE_NO_HARDWARE_RESPONSE);