[PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Michael Brown]

Marcelo, Al,
	Below is an updated patch to add SMBIOS information to /proc/smbios.
Updates have been made per Al's previous comments. Please apply.

For reference, here are previous postings:
Previous 2.4 thread:
http://marc.theaimsgroup.com/?t=108321757100001&r=1&w=1

Previous 2.6 thread:
http://marc.theaimsgroup.com/?t=108311959700002&r=1&w=1

Thank you,
Michael Brown


# This is a BitKeeper generated diff -Nru style patch.
#
# ChangeSet
#   2004/04/29 21:11:19-05:00 michael_e_brown@meb-laptop.michaels-house.net 
#   Updates per Al Viro: 
#   	-- remove MOD_{INC,DEC}_USE_COUNT
#   	-- use normal read call instead of proc read call
# 
# drivers/char/smbios.c
#   2004/04/29 21:11:17-05:00 michael_e_brown@meb-laptop.michaels-house.net +35 -47
#   Updates per comments from Al Viro.
# 
# ChangeSet
#   2004/04/29 00:16:35-05:00 michael_e_brown@meb-laptop.michaels-house.net 
#   rename smbios to proper smbios.c
# 
# drivers/char/Makefile
#   2004/04/29 00:09:15-05:00 michael_e_brown@meb-laptop.michaels-house.net +1 -1
#   rename smbios driver to 'smbios'.
# 
# drivers/char/smbios.c
#   2004/04/29 00:08:51-05:00 michael_e_brown@meb-laptop.michaels-house.net +0 -0
#   Rename: drivers/char/procsmbios.c -> drivers/char/smbios.c
# 
# ChangeSet
#   2004/04/28 23:51:19-05:00 michael_e_brown@meb-laptop.michaels-house.net 
#   Add SMBIOS driver that exports system SMBIOS information as files in /proc/smbios.
# 
# drivers/char/smbios.h
#   2004/04/28 23:50:15-05:00 michael_e_brown@meb-laptop.michaels-house.net +53 -0
# 
# drivers/char/procsmbios.c
#   2004/04/28 23:50:15-05:00 michael_e_brown@meb-laptop.michaels-house.net +251 -0
# 
# drivers/char/smbios.h
#   2004/04/28 23:50:15-05:00 michael_e_brown@meb-laptop.michaels-house.net +0 -0
#   BitKeeper file /home/michael_e_brown/kernel/bk/subtrees/smbios-24-driver/drivers/char/smbios.h
# 
# drivers/char/procsmbios.c
#   2004/04/28 23:50:15-05:00 michael_e_brown@meb-laptop.michaels-house.net +0 -0
#   BitKeeper file /home/michael_e_brown/kernel/bk/subtrees/smbios-24-driver/drivers/char/procsmbios.c
# 
# drivers/char/Makefile
#   2004/04/28 23:50:15-05:00 michael_e_brown@meb-laptop.michaels-house.net +1 -0
#   Add SMBIOS driver to build system
# 
# drivers/char/Config.in
#   2004/04/28 23:50:15-05:00 michael_e_brown@meb-laptop.michaels-house.net +1 -0
#   Add config option to turn on SMBIOS proc driver
# 
# Documentation/Configure.help
#   2004/04/28 23:50:15-05:00 michael_e_brown@meb-laptop.michaels-house.net +14 -0
#   Add help text for SMBIOS driver
# 
diff -Nru a/Documentation/Configure.help b/Documentation/Configure.help
--- a/Documentation/Configure.help	Thu Apr 29 21:13:45 2004
+++ b/Documentation/Configure.help	Thu Apr 29 21:13:45 2004
@@ -19914,6 +19914,20 @@
 
   If unsure, say N.
 
+SMBIOS table information in /proc
+CONFIG_SMBIOS
+  This driver creates two files in /proc, /proc/smbios/table_entry_point
+  and /proc/smbios/table. These two files contain the contents of the 
+  BIOS-generated SMBIOS tables. Generally, PC-like architectures made
+  after 1997 have this, it is safe to enable on any system.
+  
+  To compile this driver as a module ( = code which can be inserted in
+  and removed from the running kernel whenever you want), say M here
+  and read <file:Documentation/modules.txt>. The module will be called
+  smbios.o.
+  
+  If unsure, say N.
+
 Sony Vaio Programmable I/O Control Device support
 CONFIG_SONYPI
   This driver enables access to the Sony Programmable I/O Control
diff -Nru a/drivers/char/Config.in b/drivers/char/Config.in
--- a/drivers/char/Config.in	Thu Apr 29 21:13:45 2004
+++ b/drivers/char/Config.in	Thu Apr 29 21:13:45 2004
@@ -333,6 +333,7 @@
 if [ "$CONFIG_EXPERIMENTAL" = "y" -a "$CONFIG_X86" = "y" -a "$CONFIG_X86_64" != "y" ]; then
    dep_tristate 'Sony Vaio Programmable I/O Control Device support (EXPERIMENTAL)' CONFIG_SONYPI $CONFIG_PCI
 fi
+tristate 'SMBIOS table information in /proc' CONFIG_SMBIOS
 
 mainmenu_option next_comment
 comment 'Ftape, the floppy tape device driver'
diff -Nru a/drivers/char/Makefile b/drivers/char/Makefile
--- a/drivers/char/Makefile	Thu Apr 29 21:13:45 2004
+++ b/drivers/char/Makefile	Thu Apr 29 21:13:45 2004
@@ -265,6 +265,7 @@
 obj-$(CONFIG_HW_RANDOM) += hw_random.o
 obj-$(CONFIG_AMD_PM768) += amd76x_pm.o
 obj-$(CONFIG_BRIQ_PANEL) += briq_panel.o
+obj-$(CONFIG_SMBIOS) += smbios.o
 
 obj-$(CONFIG_ITE_GPIO) += ite_gpio.o
 obj-$(CONFIG_AU1X00_GPIO) += au1000_gpio.o
diff -Nru a/drivers/char/smbios.c b/drivers/char/smbios.c
--- /dev/null	Wed Dec 31 16:00:00 1969
+++ b/drivers/char/smbios.c	Thu Apr 29 21:13:45 2004
@@ -0,0 +1,239 @@
+/*
+ * linux/drivers/firmware/smbios.c
+ *  Copyright (C) 2002, 2003, 2004 Dell Inc.
+ *  by Michael Brown <Michael_E_Brown@dell.com>
+ *  vim:noet:ts=8:sw=8:filetype=c:textwidth=80:
+ *
+ * BIOS SMBIOS Table access
+ * conformant to DMTF SMBIOS definition
+ *   at http://www.dmtf.org/standards/smbios
+ *
+ * This code takes information provided by SMBIOS tables
+ * and presents it in procfs as:
+ *    /proc/smbios
+ *		|--> /table_entry_point
+ *		|--> /table
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License v2.0 as published by
+ * the Free Software Foundation
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ */
+
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/proc_fs.h>
+#include <asm/io.h>
+#include <asm/uaccess.h>
+#include "smbios.h"
+
+#define SMBIOS_VERSION "1.0"
+#define SMBIOS_DATE    "2004-04-29"
+
+MODULE_AUTHOR("Michael Brown <Michael_E_Brown@Dell.com>");
+MODULE_DESCRIPTION("procfs interface to SMBIOS information");
+MODULE_LICENSE("GPL");
+EXPORT_NO_SYMBOLS;
+
+struct smbios_device {
+	struct smbios_table_entry_point table_eps;
+	unsigned int smbios_table_real_length;
+};
+
+/* there shall be only one */
+static struct smbios_device the_smbios_device;
+
+static struct proc_dir_entry *smbios_dir, *table_eps_file, *table_file;
+
+static __init int
+checksum_eps(struct smbios_table_entry_point *table_eps)
+{
+	u8 *p = (u8 *)table_eps;
+	u8 checksum = 0;
+	int i=0;
+	for (i=0; i < table_eps->eps_length && i < sizeof(*table_eps); ++i) {
+		checksum += p[i];
+	}
+	return(checksum == 0);
+}
+
+static __init int
+find_table_entry_point(struct smbios_device *sdev)
+{
+	struct smbios_table_entry_point *table_eps = &(sdev->table_eps);
+	u32 fp = 0xF0000;
+	while (fp < 0xFFFFF) {
+		isa_memcpy_fromio(table_eps, fp, sizeof(*table_eps));
+		if (memcmp(table_eps->anchor, "_SM_", 4)==0 &&
+					checksum_eps(table_eps)) {
+			return 0;
+		}
+		fp += 16;
+	}
+
+	printk(KERN_INFO "SMBIOS table entry point not found in "
+			"0xF0000 - 0xFFFFF\n");
+	return -ENODEV;
+}
+
+static __init int
+find_table_max_address(struct smbios_device *sdev)
+{
+	/* break out on one of three conditions:
+	 *   -- hit table_eps.table_length
+	 *   -- hit number of items that table claims we have
+	 *   -- hit structure type 127
+	 */
+
+	u8 *buf = ioremap(sdev->table_eps.table_address,
+			sdev->table_eps.table_length);
+	u8 *ptr = buf;
+	int count = 0, keep_going = 1;
+	int max_count = sdev->table_eps.table_num_structs;
+	int max_length = sdev->table_eps.table_length;
+	while(keep_going && ((ptr - buf) <= max_length) && count < max_count){
+		if (ptr[0] == 0x7F)   /* ptr[0] is type */
+			keep_going = 0;
+
+		ptr += ptr[1]; /* ptr[1] is length, skip structure */
+		/* skip strings at end of structure */
+		while((ptr-buf) < max_length && (ptr[0] || ptr[1]))
+			++ptr;
+
+		/* string area ends in double-null. skip it. */
+		ptr += 2;
+		++count;
+	}
+	sdev->smbios_table_real_length = (ptr - buf);
+	iounmap(buf);
+
+	if (count != max_count)
+		printk(KERN_INFO "Warning: SMBIOS table structure count"
+				" does not match count specified in the"
+				" table entry point.\n"
+				" Table entry point count: %d\n"
+				" Actual count: %d\n",
+				max_count, count);
+
+	if (keep_going != 0)
+		printk(KERN_INFO "Warning: SMBIOS table does not end with a"
+				" structure type 127. This may indicate a"
+				" truncated table.");
+
+	if (sdev->smbios_table_real_length != max_length)
+		printk(KERN_INFO "Warning: BIOS specified SMBIOS table length"
+				" does not match calculated length.\n"
+				" BIOS specified: %d\n"
+				" calculated length: %d\n",
+				max_length, sdev->smbios_table_real_length);
+
+	return sdev->smbios_table_real_length;
+}
+
+/* simple procfs style. Print the whole thing and let core
+ * handle splitting it out for userspace and setting eof.
+ */
+static ssize_t
+smbios_read_table_entry_point(char *page, char **start,
+				off_t off, int count,
+				int *eof, void *data)
+{
+	unsigned int max_off = sizeof(the_smbios_device.table_eps);
+	memcpy(page, &the_smbios_device.table_eps, max_off);
+	return max_off;
+}
+
+static ssize_t smbios_read_table(struct file *file, char *buf,
+                size_t count, loff_t *ppos)
+{
+	unsigned long origppos = *ppos;
+	unsigned long max_off = the_smbios_device.smbios_table_real_length;
+	u8 *ptr;
+
+	if(*ppos >= max_off)
+		return 0;
+
+	if(count > (max_off - *ppos))
+		count = max_off - *ppos;
+
+	ptr = ioremap(the_smbios_device.table_eps.table_address, max_off);
+	if (ptr == NULL)
+		return -ENXIO;
+
+	while (*ppos < max_off) {
+		put_user(readb(ptr + *ppos), buf);
+		++(*ppos); ++buf;
+	}
+	iounmap(ptr);
+	return *ppos - origppos;
+}
+
+static struct file_operations proc_smbios_table_operations = {
+	.read   = smbios_read_table,
+};
+
+static int __init
+smbios_init(void)
+{
+	int rc=0;
+
+	printk(KERN_INFO "SMBIOS facility v%s %s\n",
+			SMBIOS_VERSION, SMBIOS_DATE);
+
+	rc = find_table_entry_point(&the_smbios_device);
+	if (rc)
+		return rc;
+
+	find_table_max_address(&the_smbios_device);
+
+	rc = -ENOMEM;
+	smbios_dir = proc_mkdir("smbios", NULL);
+	if (smbios_dir == NULL)
+		goto out;
+
+	smbios_dir->owner = THIS_MODULE;
+
+	table_eps_file = create_proc_read_entry("table_entry_point",
+						0444, smbios_dir,
+						smbios_read_table_entry_point,
+						NULL);
+	if (table_eps_file == NULL)
+		goto no_table_eps_file;
+
+	table_eps_file->owner = THIS_MODULE;
+
+	table_file = create_proc_entry("table", 0444, smbios_dir);
+	if (table_file == NULL)
+		goto no_table_file;
+
+	table_file->proc_fops = &proc_smbios_table_operations;
+	table_file->owner = THIS_MODULE;
+
+	rc = 0;
+	goto out;
+no_table_file:
+	remove_proc_entry("table_entry_point", smbios_dir);
+
+no_table_eps_file:
+	remove_proc_entry("smbios", NULL);
+
+out:
+	return rc;
+}
+
+static void __exit
+smbios_exit(void)
+{
+	remove_proc_entry("table_entry_point", smbios_dir);
+	remove_proc_entry("table", smbios_dir);
+	remove_proc_entry("smbios", NULL);
+}
+
+module_init(smbios_init);
+module_exit(smbios_exit);
diff -Nru a/drivers/char/smbios.h b/drivers/char/smbios.h
--- /dev/null	Wed Dec 31 16:00:00 1969
+++ b/drivers/char/smbios.h	Thu Apr 29 21:13:45 2004
@@ -0,0 +1,53 @@
+/*
+ * linux/drivers/firmware/smbios.c
+ *  Copyright (C) 2002, 2003, 2004 Dell Inc.
+ *  by Michael Brown <Michael_E_Brown@dell.com>
+ *  vim:noet:ts=8:sw=8:filetype=c:textwidth=80:
+ *
+ * BIOS SMBIOS Table access 
+ * conformant to DMTF SMBIOS definition
+ *   at http://www.dmtf.org/standards/smbios
+ *
+ * This code takes information provided by SMBIOS tables
+ * and presents it in sysfs.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License v2.0 as published by
+ * the Free Software Foundation
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ */
+
+#ifndef _LINUX_SMBIOS_H
+#define _LINUX_SMBIOS_H
+
+#include <linux/types.h>
+
+struct smbios_table_entry_point {
+	u8 anchor[4];
+	u8 checksum;
+	u8 eps_length;
+	u8 major_ver;
+	u8 minor_ver;
+	u16 max_struct_size;
+	u8 revision;
+	u8 formatted_area[5];
+	u8 dmi_anchor[5];
+	u8 intermediate_checksum;
+	u16 table_length;
+	u32 table_address;
+	u16 table_num_structs;
+	u8 smbios_bcd_revision;
+} __attribute__ ((packed));
+
+struct smbios_structure_header {
+	u8 type;
+	u8 length;
+	u16 handle;
+} __attribute__ ((packed));
+
+#endif				/* _LINUX_SMBIOS_H */

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[viro]

On Thu, Apr 29, 2004 at 09:21:52PM -0500, Michael Brown wrote:
> +	u32 fp = 0xF0000;
> +	while (fp < 0xFFFFF) {
> +		isa_memcpy_fromio(table_eps, fp, sizeof(*table_eps));
> +		if (memcmp(table_eps->anchor, "_SM_", 4)==0 &&
> +					checksum_eps(table_eps)) {
> +			return 0;
> +		}
> +		fp += 16;
> +	}

Stilistic note:
	for (fp = 0xf0000; fp < 0xfffff; fp += 16) {
		isa_memcpy_fromio(table_eps, fp, sizeof(*table_eps));
		if (memcmp(table_eps->anchor, "_SM_", 4) != 0)
			continue;
		if (checksum_eps(table_eps))
			return 0;
	}

> +	while(keep_going && ((ptr - buf) <= max_length) && count < max_count){
> +		if (ptr[0] == 0x7F)   /* ptr[0] is type */
> +			keep_going = 0;
> +
> +		ptr += ptr[1]; /* ptr[1] is length, skip structure */
> +		/* skip strings at end of structure */
> +		while((ptr-buf) < max_length && (ptr[0] || ptr[1]))
> +			++ptr;

It looks like an off-by-one - if ptr reaches buf + max_length - 1, ptr[1]
appears to be beyond the area it's OK to dereference.

> +                size_t count, loff_t *ppos)
> +{
> +	unsigned long origppos = *ppos;
> +	unsigned long max_off = the_smbios_device.smbios_table_real_length;
> +	u8 *ptr;
> +
> +	if(*ppos >= max_off)
> +		return 0;

Note that *ppos is signed here.  llseek() to negative and you've got a problem.

> +	while (*ppos < max_off) {
> +		put_user(readb(ptr + *ppos), buf);
> +		++(*ppos); ++buf;
> +	}

Eeek...

	a) that's called copy_to_user()
	b) you'd better check the return value (either of put_user() or
copy_to_user()).

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Michael Brown]

Good stuff! Thanks for the feedback Al. Give me a few minutes and I will
send an updated patch.

Comments/Questions below.

On Thu, 2004-04-29 at 22:34, viro@parcelfarce.linux.theplanet.co.uk
wrote:
> On Thu, Apr 29, 2004 at 09:21:52PM -0500, Michael Brown wrote:
> > +	u32 fp = 0xF0000;
> > +	while (fp < 0xFFFFF) {
> > +		isa_memcpy_fromio(table_eps, fp, sizeof(*table_eps));
> > +		if (memcmp(table_eps->anchor, "_SM_", 4)==0 &&
> > +					checksum_eps(table_eps)) {
> > +			return 0;
> > +		}
> > +		fp += 16;
> > +	}
> 
> Stilistic note:
> 	for (fp = 0xf0000; fp < 0xfffff; fp += 16) {
> 		isa_memcpy_fromio(table_eps, fp, sizeof(*table_eps));
> 		if (memcmp(table_eps->anchor, "_SM_", 4) != 0)
> 			continue;
> 		if (checksum_eps(table_eps))
> 			return 0;
> 	}

Will change. I like your version.

Originally copied from Alan Cox's stuff, so if Alan's style is off, oh
well... :-)

> 
> > +	while(keep_going && ((ptr - buf) <= max_length) && count < max_count){
> > +		if (ptr[0] == 0x7F)   /* ptr[0] is type */
> > +			keep_going = 0;
> > +
> > +		ptr += ptr[1]; /* ptr[1] is length, skip structure */
> > +		/* skip strings at end of structure */
> > +		while((ptr-buf) < max_length && (ptr[0] || ptr[1]))
> > +			++ptr;
> 
> It looks like an off-by-one - if ptr reaches buf + max_length - 1, ptr[1]
> appears to be beyond the area it's OK to dereference.

Great spot. Updating, changed "<= max_length" to "<= (max_length-1)".

> 
> > +                size_t count, loff_t *ppos)
> > +{
> > +	unsigned long origppos = *ppos;
> > +	unsigned long max_off = the_smbios_device.smbios_table_real_length;
> > +	u8 *ptr;
> > +
> > +	if(*ppos >= max_off)
> > +		return 0;
> 
> Note that *ppos is signed here.  llseek() to negative and you've got a problem.

Added "|| *ppos < 0" to the check.

> 
> > +	while (*ppos < max_off) {
> > +		put_user(readb(ptr + *ppos), buf);
> > +		++(*ppos); ++buf;
> > +	}
> 
> Eeek...
> 
> 	a) that's called copy_to_user()
> 	b) you'd better check the return value (either of put_user() or
> copy_to_user()).

Ok, will update, but I have one question. for (A), is this equivalent to
copy_to_user() even with the readb() in there? Sorry if this is a stupid
question.

If I get a bad return from either of these, is "return -EINVAL"
appropriate?
--
Michael

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Marcelo Tosatti]

On Thu, Apr 29, 2004 at 09:21:52PM -0500, Michael Brown wrote:
> Marcelo, Al,
> 	Below is an updated patch to add SMBIOS information to /proc/smbios.
> Updates have been made per Al's previous comments. Please apply.
> 
> For reference, here are previous postings:
> Previous 2.4 thread:
> http://marc.theaimsgroup.com/?t=108321757100001&r=1&w=1
> 
> Previous 2.6 thread:
> http://marc.theaimsgroup.com/?t=108311959700002&r=1&w=1

Hi Michael,

Has your patch been accepted into v2.6? 

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Andi Kleen]

Michael Brown <mebrown@michaels-house.net> writes:

> 	Below is an updated patch to add SMBIOS information to /proc/smbios.
> Updates have been made per Al's previous comments. Please apply.

What is this good for? There are tools to read this from
/dev/mem; and that is fine because the information is static.
There is no reason to bloat the kernel with this.

-Andi

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Michael Brown]

On Fri, 2004-04-30 at 14:22, Andi Kleen wrote:
> Michael Brown <mebrown@michaels-house.net> writes:
> 
> > 	Below is an updated patch to add SMBIOS information to /proc/smbios.
> > Updates have been made per Al's previous comments. Please apply.
> 
> What is this good for? There are tools to read this from
> /dev/mem; and that is fine because the information is static.
> There is no reason to bloat the kernel with this.

As I mentioned in my first posting of this to l-k, there are three
reasons why this driver is necessary:

-- This information is, in the very near future, _not_ going to be
static anymore. There will be systems that update the information in
dynamically during SMIs.

-- SMBIOS consists of two things, the table entry point and the table
itself. The table entry point is always in 0xF0000 - 0xFFFFF.
Traditionally, the actual table has been here as well. BIOS is running
out of space here and future systems are moving this information to high
memory. /dev/mem will not allow access to memory above top of system
RAM.

-- Red Hat has a /dev/mem patch in their tree that restricts access to
RAM above 1MB. 

Because of all of these reasons, we feel it is a good thing to have a
stable method to get to the SMBIOS information that will work into the
future. Our userspace libs will try to use this driver to access SMBIOS,
but fall back to /dev/mem if this driver is not available. (with the
caveat that nothing will work if table >1MB and this driver not
present.)

As for the "bloat" argument: this driver is about the most trivial
driver I can conceive of that does useful work. It is 250 raw lines of
code, comparable in size to /proc/meminfo or /proc/cpuinfo. 

This 250 line driver allows us to move a few thousand lines of code from
Dell's current, proprietary systems management driver into userspace. If
this approach is accepted, I am pushing to work on opening up other
pieces of Dell's current proprietary drivers. This work is a
proof-of-concept to management that this approach can work.
--
Michael

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Michael Brown]

On Fri, 2004-04-30 at 22:30, Michael Brown wrote:
> On Fri, 2004-04-30 at 14:22, Andi Kleen wrote:
> > Michael Brown <mebrown@michaels-house.net> writes:
> > 
> > > 	Below is an updated patch to add SMBIOS information to /proc/smbios.
> > > Updates have been made per Al's previous comments. Please apply.
> > 
> > What is this good for? There are tools to read this from
> > /dev/mem; and that is fine because the information is static.
> > There is no reason to bloat the kernel with this.

Sorry, I missed one other reason. 

This procfs/sysfs driver allows access to smbios information by
non-root, non CAP_SYS_RAWIO users. I've had several occasions where I
have been bitten by having to be root to read smbios when I did not need
root for anything else.
--
Michael

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Andi Kleen]

> -- This information is, in the very near future, _not_ going to be
> static anymore. There will be systems that update the information in
> dynamically during SMIs.

That's fine - /dev/mem can handle that too. An user will have to
poll for changes anyways, so having it it /proc does not have
any advantages.

I would see the sense if there was an interrupt to notify the
the system of changes, but there isn't ;-)

> 
> -- SMBIOS consists of two things, the table entry point and the table
> itself. The table entry point is always in 0xF0000 - 0xFFFFF.
> Traditionally, the actual table has been here as well. BIOS is running
> out of space here and future systems are moving this information to high
> memory. /dev/mem will not allow access to memory above top of system
> RAM.

mmap /dev/mem of high memory should work. Did you try it?
For read/write it would be fairly easy to add, just needs a kmap.

> -- Red Hat has a /dev/mem patch in their tree that restricts access to
> RAM above 1MB. 

That's their breakage.

Clearly it is a useless change anyways because you can easily circumvent
it by just accessing any bus master hardware in the system.

> This procfs/sysfs driver allows access to smbios information by
> non-root, non CAP_SYS_RAWIO users. I've had several occasions where I
> have been bitten by having to be root to read smbios when I did not need
> root for anything else.

That is what suid root was invented for. It is a fairly nifty mechanism,
did you ever try it?  I know that some sysadmins try to get away from
it, but that is clearly misguided because it is not any more dangerous than
adding more kernel code.

-Andi

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Lev Makhlis]

> > -- This information is, in the very near future, _not_ going to be
> > static anymore. There will be systems that update the information in
> > dynamically during SMIs.
>
> That's fine - /dev/mem can handle that too. An user will have to
> poll for changes anyways, so having it it /proc does not have
> any advantages.

One problem is that /dev/mem access isn't atomic.  You need to read
a pointer, then follow the pointer to read data.  If the pointer changes
in the middle, you lose.  That said, I don't see any mechanism that
helps avoid that in kernel, either.

Re: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Andi Kleen]

On Sat, May 01, 2004 at 04:00:44PM -0500, Lev Makhlis wrote:
> > > -- This information is, in the very near future, _not_ going to be
> > > static anymore. There will be systems that update the information in
> > > dynamically during SMIs.
> >
> > That's fine - /dev/mem can handle that too. An user will have to
> > poll for changes anyways, so having it it /proc does not have
> > any advantages.
> 
> One problem is that /dev/mem access isn't atomic.  You need to read
> a pointer, then follow the pointer to read data.  If the pointer changes
> in the middle, you lose.  That said, I don't see any mechanism that
> helps avoid that in kernel, either.
> 

Exactly. The kernel code doesn't help with that.

You could always read everything twice and compare and do that until
it stops changing. I personally think it is overkill, because SMBIOS
information is normally only used for diagnosis, where a rare race
is not that bad.

Also hopefully the SMI handlers will change the thing in a atomic way,
so this may never happen.

-Andi

RE: [PATCH 2.4] add SMBIOS information to /proc/smbios -- UPDATED

[Michael_E_Brown]

Yes, Greg KH has pushed it into current -mm, and intends to push it up
for 2.6.7.
--
Michael

> -----Original Message-----
> From: Marcelo Tosatti [mailto:marcelo.tosatti@cyclades.com] 
> Sent: Monday, May 03, 2004 6:40 PM
> To: Michael Brown
> Cc: linux-kernel@vger.kernel.org; Brown, Michael E; 
> viro@parcelfarce.linux.theplanet.co.uk
> Subject: Re: [PATCH 2.4] add SMBIOS information to 
> /proc/smbios -- UPDATED
> 
> 
> On Thu, Apr 29, 2004 at 09:21:52PM -0500, Michael Brown wrote:
> > Marcelo, Al,
> > 	Below is an updated patch to add SMBIOS information to 
> /proc/smbios.
> > Updates have been made per Al's previous comments. Please apply.
> > 
> > For reference, here are previous postings:
> > Previous 2.4 thread:
> > http://marc.theaimsgroup.com/?t=108321757100001&r=1&w=1
> > 
> > Previous 2.6 thread:
> > http://marc.theaimsgroup.com/?t=108311959700002&r=1&w=1
> 
> Hi Michael,
> 
> Has your patch been accepted into v2.6? 
>