On Fri, 17 Oct 2008, Zhao Yakui wrote: > On Thu, 2008-10-16 at 20:08 -0400, Len Brown wrote: > > Yakui, > > I think your patch below from > > http://bugzilla.kernel.org/show_bug.cgi?id=11481 > > is close, but not quite right. > > > > "reboot=a" should not be necessary, as reboot_type = BOOT_ACPI > > by default, yes? > the default reboot_type is BOOT_KBD. So it is necessary to add the boot > option of "reboot=a" to use the ACPI reboot mechanism. hmm, arch/x86/kernel/reboot.c has this: /* * Keyboard reset and triple fault may result in INIT, not RESET, which * doesn't work when we're in vmx root mode. Try ACPI first. */ enum reboot_type reboot_type = BOOT_ACPI; Am I mis-reading the code? > > Re: DMI to ignore a missing (acpi_gbl_FADT.flags & > > ACPI_FADT_RESET_REGISTER).... > > > > If we need DMI, > > it should be checked at _boot_ time, not at _reboot_ > > time, and the table should be marked _initdata. > Ok. I will refresh it. The DMI table will be put into _initdata > section. I'm okay with a DMI entry now, but only if we work on figuring out how to delete the DMI entries to fix the problem for all system w/o using DMI. > > But the real question is if we should be checking this bit > > at all. The fact that we are adding DMI to ignore it suggest > > that we should think about ignoring it by default, and instead > > perhaps sanity checking acpi_gbl_FADT.reset_register > > and basing our decision to use it on that? > According to the ACPI spec the bit indicates whether the ACPI reset > mechanism is supported. If it is zero, maybe OS should not use the ACPI > reset mechanism even when ACPI_RESET_REG is defined. understood. > On some laptops the system can't be rebooted by using default > reboot_type(BOOT_KBD) after it is resumed from S3. If the ACPI reboot is > used, the system can be rebooted. But unfortunately the FADT flag bit > indicates that ACPI reboot mechanism is not supported. This seems > contradictory. This contradiction suggests that Windows is not following the ACPI spec, and is not checking the FADT bit before using the ACPI reboot mechanism. > In the patch the FADT flag bit is checked to comply with the ACPI spec. > But in order to use the ACPI reboot mechanism on some laptops, the DMI > check table is added in which the FADT flag bit is ignored. Understood. The problem is that it is very likely that Windows has NO DMI entry for these laptops, yet it still works. So if we can't figure out why these laptops work with Windows, we are setting ourselves up to create an unmaintainable DMI list, possibly including a very large number of systems which have this BIOS bug. > Of course what you said is also OK. When the system is expected to be > rebooted using the ACPI reboot mechanism, the FADT flag bit is ignored > and instead reset_register/reset_value is checked. If so, the DMI check > table is redundant. Redundant = deleted. If we can figure out how/when to use the ACPI reset register w/o consulting the FADT bit, then there should be no DMI table. thanks, -Len -- To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
