linux-kernel - 2.6.18 suspend regression on Intel Macs

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Date:	Mon, 09 Oct 2006 20:19:41 +0200
From:	Frédéric Riss <frederic.riss@...il.com>
To:	Linux Kernel list <linux-kernel@...r.kernel.org>
Cc:	Linus Torvalds <torvalds@...l.org>, len.brown@...el.com,
	Pavel Machek <pavel@....cz>
Subject: 2.6.18 suspend regression on Intel Macs

Well, I'm not sure it qualifies as a regression, because AFAIK no
official kernels can s2ram/resume Intel Macs correctly out of the box.

There has already been some discussion about the SCI_EN ACPI control bit
not being set when the Mactel boxes come out of suspend to ram. 
http://marc.theaimsgroup.com/?l=linux-acpi&m=114957637501557&w=2
http://marc.theaimsgroup.com/?l=linux-kernel&m=115005083610700&w=2
http://bugme.osdl.org/show_bug.cgi?id=6670

The symptom is:
    irq 9: nobody cared (try booting with the "irqpoll" option)
    Disabling IRQ #9
when the system comes out of sleep, making ACPI non-functional.

Two days after having released 2.6.17, Linus commited a fix for this
issue in his tree (commit 5603509137940f4cbc577281cee62110d4097b1b):

@@ -812,6 +812,9 @@ static int irqrouter_resume(struct sys_d

ACPI_FUNCTION_TRACE("irqrouter_resume");

+ /* Make sure SCI is enabled again (Apple firmware bug?) */
+ acpi_set_register(ACPI_BITREG_SCI_ENABLE, 1, ACPI_MTX_DO_NOT_LOCK);
+

I since then used lightly patched 2.6.17 kernels on my MacMini without a
problem. With the release of 2.6.18, I decided to switch to a vanilla
kernel but I realized that the above issue reappeared. 

I tracked it down to the ACPI merge that took place on July 1st. More
precisely the commit 967440e3be1af06ad4dc7bb18d2e3c16130fe067 (ACPI:
ACPICA 20060623) contains the following hunk:

@@ -635,6 +663,25 @@ acpi_status acpi_hw_register_write(u8 us

 	case ACPI_REGISTER_PM1_CONTROL:	/* 16-bit access */

+		/*
+		 * Perform a read first to preserve certain bits (per ACPI spec)
+		 *
+		 * Note: This includes SCI_EN, we never want to change this bit
+		 */
+		status = acpi_hw_register_read(ACPI_MTX_DO_NOT_LOCK,
+					       ACPI_REGISTER_PM1_CONTROL,
+					       &read_value);
+		if (ACPI_FAILURE(status)) {
+			goto unlock_and_exit;
+		}
+
+		/* Insert the bits to be preserved */
+
+		ACPI_INSERT_BITS(value, ACPI_PM1_CONTROL_PRESERVED_BITS,
+				 read_value);
+
+		/* Now we can write the data */
+
 		status =
 		    acpi_hw_low_level_write(16, value,
 					    &acpi_gbl_FADT->xpm1a_cnt_blk);

which makes Linus' fix a no-op, because it disallows setting the SCI_EN
bit.

Fred.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/