public inbox for linux-ia64@vger.kernel.org
 help / color / mirror / Atom feed
From: Bjorn Helgaas <bjorn.helgaas@hp.com>
To: linux-ia64@vger.kernel.org
Subject: [RFC] prevent "dd if=/dev/mem" crash
Date: Fri, 17 Oct 2003 22:10:36 +0000	[thread overview]
Message-ID: <marc-linux-ia64-106642876514553@msgid-missing> (raw)

This is the generic part of a change to prevent "dd if=/dev/mem"
from causing a machine check on ia64.

read_mem() and write_mem() already check the requested address
against "high_memory", but that is only a complete check if
everything from 0 to high_memory is valid, readable/writable
memory.  Obviously that's not the case for architectures with
discontiguous memory, like ia64.

Old behavior:

    # dd if=/dev/mem of=/dev/null
    <unrecoverable machine check>

New behavior (this system has a hole from 0-16MB, then memory
from 16MB-1GB):

    # dd if=/dev/mem of=/dev/null
    0+0 records in
    0+0 records out
    0 bytes transferred in 0.000282 seconds (0 bytes/sec)

    # dd if=/dev/mem of=/dev/null bs=1M skip\x16 
    1004+10 records in
    1004+10 records out
    1056964608 bytes transferred in 1.629262 seconds (648738280 bytes/sec)

I expect there are probably different opinions about the idea
that "dd if=/dev/mem" exits without doing anything.  Sparc and
68K have nearby code that bit-buckets writes and returns zeroes
for reads of page zero.  We could do that, too, but it seems like
kind of a hack, and holes on ia64 can be BIG (on the order of
256GB for one box).

So flame away :-)

The patch below is mangled so it won't apply easily.  If this
seems a reasonable approach, I'll submit the ia64 piece first,
then repost this.

Bjorn

=== drivers/char/mem.c 1.44 vs edited ==--- 1.44/ drivers/char/mem.c	Sun Sep 21 15:50:34 2003
+++ edited/ drivers/char/mem.c	Fri Oct 17 15:37:47 2003
@@ -79,6 +79,24 @@
 #endif
 }
 
+static inline int valid_mem_range(unsigned long addr, size_t *count)
+{
+#if defined(CONFIG_IA64)
+	return efi_valid_mem_range(addr, count);
+#else
+	unsigned long end_mem;
+
+	end_mem = __pa(high_memory);
+	if (addr >= end_mem)
+		return 0;
+
+	if (*count > end_mem - addr)
+		*count = end_mem - addr;
+
+	return 1;
+#endif
+}
+
 static ssize_t do_write_mem(struct file * file, void *p, unsigned long realp,
 			    const char * buf, size_t count, loff_t *ppos)
 {
@@ -113,14 +131,10 @@
 			size_t count, loff_t *ppos)
 {
 	unsigned long p = *ppos;
-	unsigned long end_mem;
 	ssize_t read;
 
-	end_mem = __pa(high_memory);
-	if (p >= end_mem)
+	if (!valid_mem_range(p, &count))
 		return 0;
-	if (count > end_mem - p)
-		count = end_mem - p;
 	read = 0;
 #if defined(__sparc__) || (defined(__mc68000__) && defined(CONFIG_MMU))
 	/* we don't have page 0 mapped on sparc and m68k.. */
@@ -149,13 +163,9 @@
 			 size_t count, loff_t *ppos)
 {
 	unsigned long p = *ppos;
-	unsigned long end_mem;
 
-	end_mem = __pa(high_memory);
-	if (p >= end_mem)
+	if (!valid_mem_range(p, &count))
 		return 0;
-	if (count > end_mem - p)
-		count = end_mem - p;
 	return do_write_mem(file, __va(p), p, buf, count, ppos);
 }
 
=== include/linux/efi.h 1.3 vs edited ==--- 1.3/ include/linux/efi.h	Thu Aug  7 14:01:48 2003
+++ edited/ include/linux/efi.h	Thu Oct 16 16:54:52 2003
@@ -266,6 +266,7 @@
 extern u64 efi_get_iobase (void);
 extern u32 efi_mem_type (unsigned long phys_addr);
 extern u64 efi_mem_attributes (unsigned long phys_addr);
+extern int efi_valid_mem_range (unsigned long phys_addr, unsigned long *count);
 
 /*
  * Variable Attributes


             reply	other threads:[~2003-10-17 22:10 UTC|newest]

Thread overview: 24+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2003-10-17 22:10 Bjorn Helgaas [this message]
2003-10-17 22:19 ` [RFC] prevent "dd if=/dev/mem" crash Luck, Tony
2003-10-17 22:23 ` Matt Mackall
2003-10-17 22:40 ` Andreas Schwab
2003-10-17 22:50 ` Andrew Morton
2003-10-17 23:25 ` Bjorn Helgaas
2003-10-17 23:55 ` Andrew Morton
2003-10-18  0:15 ` William Lee Irwin III
2003-10-18  0:21 ` David Mosberger
2003-10-18  0:49 ` Andrew Morton
2003-10-18  1:31 ` Matt Chapman
2003-10-18  1:41 ` Andrew Morton
2003-10-18  1:48 ` David Mosberger
2003-10-18  2:01 ` Andrew Morton
2003-10-18  2:01 ` Matt Chapman
2003-10-19 11:25 ` Eric W. Biederman
2003-10-19 18:17 ` Pavel Machek
2003-10-19 19:01 ` William Lee Irwin III
2003-10-20 15:17 ` Bjorn Helgaas
2003-10-20 17:42 ` Bjorn Helgaas
2003-10-20 18:48 ` David Mosberger
2003-10-23  8:33 ` Martin Pool
2003-10-23  9:31 ` Zoltan Menyhart
2003-10-23 21:05 ` Bjorn Helgaas

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=marc-linux-ia64-106642876514553@msgid-missing \
    --to=bjorn.helgaas@hp.com \
    --cc=linux-ia64@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox