Skip to contentSkip to conversation

Don't allow MICROCODE built as a modul

PATCH In review S.Çağlar Onur 3 participants 5 messages Oct 17, 2006 – Oct 18, 2006

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Don't allow MICROCODE built as a modul

Hi;

Attached patch [against xen-3.0.3-testing.hg]
fixes "/lib/modules/2.6.12.6-xen/kernel/arch/xen/i386/kernel/microcode.ko needs unknown symbol sys_munlock" problem by not allowing microcode as a module.

Cheers

Signature · 4 lines
S.Çağlar Onur <[email protected]>
http://cekirdek.pardus.org.tr/~caglar/

Linux is like living in a teepee. No Windows, no Gates and an Apache in house!
linux-2.6-xen-sparse/arch/i386/Kconfig +1 -1
diff -r b2f2b7738aa2 linux-2.6-xen-sparse/arch/i386/Kconfig
@@ -387,7 +387,7 @@ config X86_REBOOTFIXUPS
 	  Say N otherwise.
 
 config MICROCODE
-	tristate "/dev/cpu/microcode - Intel IA32 CPU microcode support"
+	bool "/dev/cpu/microcode - Intel IA32 CPU microcode support"
 	depends on !XEN_UNPRIVILEGED_GUEST
 	---help---
 	  If you say Y here and also to "/dev file system support" in the
linux-2.6-xen-sparse/arch/x86_64/Kconfig +1 -1
diff -r b2f2b7738aa2 linux-2.6-xen-sparse/arch/x86_64/Kconfig
@@ -158,7 +158,7 @@ config X86_GOOD_APIC
 	default y
 
 config MICROCODE
-	tristate "/dev/cpu/microcode - Intel CPU microcode support"
+	bool "/dev/cpu/microcode - Intel CPU microcode support"
 	---help---
 	  If you say Y here the 'File systems' section, you will be
 	  able to update the microcode on Intel processors. You will

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#1 lore raw
Quoted 6 lines · S.Çağlar Onur

Hi;

Attached patch [against xen-3.0.3-testing.hg]
fixes "/lib/modules/2.6.12.6-xen/kernel/arch/xen/i386/kernel/microcode.ko needs unknown symbol sys_munlock" problem by not allowing microcode as a module.

That's not fixing the underlying problem. Why would the microcode driver require sys_munlock()? I'm sure one of the distros would have provided a patch already if they were seeing this problem.

-- Keir

#2 lore raw

17 Eki 2006 Sal 22:54 tarihinde, Keir Fraser şunları yazmıştı:

Quoted 3 lines

That's not fixing the underlying problem. Why would the microcode driver require sys_munlock()? I'm sure one of the distros would have provided a patch already if they were seeing this problem.

Hmm, a little google search shows SUSE already has [1], im not sure original author submits or not but here it is [2]

[1] http://lists.opensuse.org/opensuse-commit/2006-04/msg00962.html
[2] http://cekirdek.pardus.org.tr/~caglar/xen-microcode-modular

Signature · 4 lines
S.Çağlar Onur <[email protected]>
http://cekirdek.pardus.org.tr/~caglar/

Linux is like living in a teepee. No Windows, no Gates and an Apache in house!

_______________________________________________
Xen-devel mailing list
[email protected]
http://lists.xensource.com/xen-devel

#3 lore raw
Quoted 10 lines · S.Çağlar Onur

17 Eki 2006 Sal 22:54 tarihinde, Keir Fraser şunları yazmıştı:

Quoted 3 lines

That's not fixing the underlying problem. Why would the microcode driver require sys_munlock()? I'm sure one of the distros would have provided a patch already if they were seeing this problem.

Hmm, a little google search shows SUSE already has [1], im not sure original author submits or not but here it is [2]

[1] http://lists.opensuse.org/opensuse-commit/2006-04/msg00962.html
[2] http://cekirdek.pardus.org.tr/~caglar/xen-microcode-modular

Ah yes, that looks sensible. Perhaps if we ask nicely Jan will post it to the list with a signed-off-by attribution. :-)

-- Keir

#4 lore raw
Quoted 8 lines
Quoted 1 line
Quoted 1 line

Keir Fraser <[email protected]> 17.10.06 22:19 >>>

Quoted 5 lines · S.Ça*lar Onur

17 Eki 2006 Sal 22:54 tarihinde, Keir Fraser *unlar* yazm**t*:

Quoted 3 lines

That's not fixing the underlying problem. Why would the microcode driver require sys_munlock()? I'm sure one of the distros would have provided a patch already if they were seeing this problem.

Quoted 8 lines
Quoted 5 lines

Hmm, a little google search shows SUSE already has [1], im not sure original author submits or not but here it is [2]

[1] http://lists.opensuse.org/opensuse-commit/2006-04/msg00962.html
[2] http://cekirdek.pardus.org.tr/~caglar/xen-microcode-modular

Ah yes, that looks sensible. Perhaps if we ask nicely Jan will post it to the list with a signed-off-by attribution. :-)

Eliminating the dependency on sys_m{,un}lock as well as the needless use of static variables.

arch/i386/kernel/microcode-xen.c +17 -23
Index: sle10-sp1-2006-10-05/arch/i386/kernel/microcode-xen.c
===================================================================
@@ -50,9 +50,6 @@ MODULE_LICENSE("GPL");
 
 /* no concurrent ->write()s are allowed on /dev/cpu/microcode */
 static DECLARE_MUTEX(microcode_sem);
-
-static void __user *user_buffer;	/* user area microcode data buffer */
-static unsigned int user_buffer_size;	/* it's size */
 				
 static int microcode_open (struct inode *unused1, struct file *unused2)
 {
@@ -60,21 +57,26 @@ static int microcode_open (struct inode 
 }
 
 
-static int do_microcode_update (void)
+static int do_microcode_update (const void __user *ubuf, size_t len)
 {
 	int err;
-	dom0_op_t op;
+	void *kbuf;
 
-	err = sys_mlock((unsigned long)user_buffer, user_buffer_size);
-	if (err != 0)
-		return err;
-
-	op.cmd = DOM0_MICROCODE;
-	set_xen_guest_handle(op.u.microcode.data, user_buffer);
-	op.u.microcode.length = user_buffer_size;
-	err = HYPERVISOR_dom0_op(&op);
+	kbuf = vmalloc(len);
+	if (!kbuf)
+		return -ENOMEM;
+
+	if (copy_from_user(kbuf, ubuf, len) == 0) {
+		dom0_op_t op;
+
+		op.cmd = DOM0_MICROCODE;
+		set_xen_guest_handle(op.u.microcode.data, kbuf);
+		op.u.microcode.length = len;
+		err = HYPERVISOR_dom0_op(&op);
+	} else
+		err = -EFAULT;
 
-	(void)sys_munlock((unsigned long)user_buffer, user_buffer_size);
+	vfree(kbuf);
 
 	return err;
 }
@@ -88,17 +90,9 @@ static ssize_t microcode_write (struct f
 		return -EINVAL;
 	}
 
-	if ((len >> PAGE_SHIFT) > num_physpages) {
-		printk(KERN_ERR "microcode: too much data (max %ld pages)\n", num_physpages);
-		return -EINVAL;
-	}
-
 	down(&microcode_sem);
 
-	user_buffer = (void __user *) buf;
-	user_buffer_size = (int) len;
-
-	ret = do_microcode_update();
+	ret = do_microcode_update(buf, len);
 	if (!ret)
 		ret = (ssize_t)len;