Date: Wed, 7 Jun 2006 15:24:04 -0400 From: Jung-uk Kim <jkim@FreeBSD.org> To: Alexander Logvinov <abuse@akavia.ru> Cc: freebsd-acpi@FreeBSD.org, "Moore, Robert" <robert.moore@intel.com> Subject: Re: Re[2]: Machine did not reboot Message-ID: <200606071524.07284.jkim@FreeBSD.org> In-Reply-To: <121000959.20060607154424@akavia.ru> References: <1182686709.20060605133201@akavia.ru> <200606062022.59336.jkim@FreeBSD.org> <121000959.20060607154424@akavia.ru>
next in thread | previous in thread | raw e-mail | index | archive | help
--Boundary-00=_XfyhEZz3nBeaHyx
Content-Type: text/plain;
charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
On Wednesday 07 June 2006 01:44 am, Alexander Logvinov wrote:
> Hello.
>
> >> > > I had FreeBSD 5.4. After entering 'shutdown -r now' the
> >> > > system hanged on: 'Shutting down ACPI'
> >> > > 'Rebooting'
> >> > > but did not reboot.
> >> > > Upgraded to 6.1, it didn't help.
> >> > > hw.acpi.disable_on_poweroff="1" has no effect. What should I
> >> > > do?
> >> > > Motherboard: Chaintech 7VJL with latest BIOS.
> >> >
> >> > Try the reset_register method. I have MFC'd the patch to
> >> > RELENG_6 so you can cvsup, recompile your acpi.ko, and test.
>
> It doesn't help. Because acpi_shutdown_final in function goes to
>
> else if (panicstr == NULL) {
> printf("Shutting down ACPI\n");
> AcpiTerminate();
> }
>
> and hangs up.
>
> >> RB_AUTOBOOT is defined as 0 in sys/reboot.h. I don't think this
> >> test will ever work:
> >> if ((howto & RB_AUTOBOOT) != 0 &&
> >> AcpiGbl_FADT->ResetRegSup) {
> >
> > It's little radical but what do you think about the attached
> > patch? I don't think we have to call AcpiTerminate() to reboot
> > at all. In fact, I have a box which does not reboot. Writing
> > ACPI_DISABLE to SMI_CMD hangs the system and it does not support
> > RESET_REG. :-( If I don't call AcpiTerminate(), everything's
> > fine.
>
> I'll try this patch soon, thanks.
I don't know what ACPI spec. says (I guess I'll have to look it up)
but Linux doesn't seem to use it except for ACPI init failure case.
Now I have another evil hack (attached). It will help rebooting
without RESET_REG support and/or with broken BIOS. Basically, it
just bypasses AcpiDisable(), which may cause hang when ACPI_DISABLE
through SMI_CMD is issued.
Thanks,
Jung-uk Kim
--Boundary-00=_XfyhEZz3nBeaHyx
Content-Type: text/plain;
charset="iso-8859-1";
name="acpi_shutdown.diff"
Content-Transfer-Encoding: 7bit
Content-Disposition: attachment;
filename="acpi_shutdown.diff"
--- sys/dev/acpica/acpi.c 6 Jun 2006 18:30:38 -0000
+++ sys/dev/acpica/acpi.c 7 Jun 2006 19:08:27 -0000
@@ -240,6 +240,12 @@
SYSCTL_INT(_debug_acpi, OID_AUTO, do_powerstate, CTLFLAG_RW,
&acpi_do_powerstate, 1, "Turn off devices when suspending.");
+/* Disable ACPI when shutting down. */
+static int acpi_do_disable = 0;
+TUNABLE_INT("debug.acpi.do_disable", &acpi_do_disable);
+SYSCTL_INT(_debug_acpi, OID_AUTO, do_disable, CTLFLAG_RW,
+ &acpi_do_disable, 0, "Disable ACPI when shutting down.");
+
/* Allow users to override quirks. */
TUNABLE_INT("debug.acpi.quirks", &acpi_quirks);
@@ -1645,7 +1651,7 @@
DELAY(1000000);
printf("ACPI power-off failed - timeout\n");
}
- } else if ((howto & RB_AUTOBOOT) != 0 && AcpiGbl_FADT->ResetRegSup) {
+ } else if ((howto & RB_HALT) == 0 && AcpiGbl_FADT->ResetRegSup) {
status = AcpiHwLowLevelWrite(
AcpiGbl_FADT->ResetRegister.RegisterBitWidth,
AcpiGbl_FADT->ResetValue, &AcpiGbl_FADT->ResetRegister);
@@ -1656,6 +1662,9 @@
printf("ACPI reset failed - timeout\n");
}
} else if (panicstr == NULL) {
+ /* XXX Evil hack to bypass AcpiDisable(). */
+ if (!acpi_do_disable)
+ AcpiGbl_OriginalMode = ACPI_SYS_MODE_ACPI;
printf("Shutting down ACPI\n");
AcpiTerminate();
}
--Boundary-00=_XfyhEZz3nBeaHyx--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200606071524.07284.jkim>