lkml.org 
[lkml]   [2006]   [Sep]   [28]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
From
Subject[PATCH] Fudge block count for 32-bit statfs() of large filesystems
Date
Recently we have noticed that apps running in compatibility mode on  
x86_64 could not statfs() the filesystem (e.g. just to check for free
space remaining). While it is true that these applications should
probably be using the 64-bit call, we have worked around the problem
with a fudge to compat.c enabling statfs() to return a larger block
size and a smaller block count.

I realise this is a horrible hack, just wanted to open a discussion
about possible alternatives! In i386 Linux the filesystem code
itself is able to make the decision to return bogus values (since it
knows the caller is 32-bit), but in x86_64 it appears that the
filesystem-specific code is unable to distinguish between 32- and 64-
bit callers.

The patch is untested, but we are currently testing a slightly
different but equivalent patch to a Redhat Enterprise Linux kernel.


--- linux-2.6.18/fs/compat.c.orig 2006-09-29 09:22:32.000000000 +1000
+++ linux-2.6.18/fs/compat.c 2006-09-29 09:32:28.000000000 +1000
@@ -164,11 +164,34 @@ asmlinkage long compat_sys_newfstat(unsi
static int put_compat_statfs(struct compat_statfs __user *ubuf,
struct kstatfs *kbuf)
{
+ /*
+ * Large filesystems often have block counts that cannot be
represented in 32 bits.
+ * Rather than returning EFAULT to 32-bit apps in compatibility
mode, we fudge
+ * the blocksize upwards to keep the block count < 2^32
+ *
+ * For now the maximum 'fudge factor' is 16, i.e. the block size
will be <<4
+ * and the block count will be >>4.
+ */
+ int block_shift = 0;

if (sizeof ubuf->f_blocks == 4) {
if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
- 0xffffffff00000000ULL)
+ 0xfffffff000000000ULL)
return -EOVERFLOW;
+
+ if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
+ 0xfffffff800000000ULL)
+ block_shift = 4;
+ else if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
+ 0xfffffffc00000000ULL)
+ block_shift = 3;
+ else if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
+ 0xfffffffe00000000ULL)
+ block_shift = 2;
+ else if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
+ 0xffffffff00000000ULL)
+ block_shift = 1;
+
/* f_files and f_ffree may be -1; it's okay
* to stuff that into 32 bits */
if (kbuf->f_files != 0xffffffffffffffffULL
@@ -180,10 +203,10 @@ static int put_compat_statfs(struct comp
}
if (!access_ok(VERIFY_WRITE, ubuf, sizeof(*ubuf)) ||
__put_user(kbuf->f_type, &ubuf->f_type) ||
- __put_user(kbuf->f_bsize, &ubuf->f_bsize) ||
- __put_user(kbuf->f_blocks, &ubuf->f_blocks) ||
- __put_user(kbuf->f_bfree, &ubuf->f_bfree) ||
- __put_user(kbuf->f_bavail, &ubuf->f_bavail) ||
+ __put_user(kbuf->f_bsize << block_shift, &ubuf->f_bsize) ||
+ __put_user(kbuf->f_blocks >> block_shift, &ubuf->f_blocks) ||
+ __put_user(kbuf->f_bfree >> block_shift, &ubuf->f_bfree) ||
+ __put_user(kbuf->f_bavail >> block_shift, &ubuf->f_bavail) ||
__put_user(kbuf->f_files, &ubuf->f_files) ||
__put_user(kbuf->f_ffree, &ubuf->f_ffree) ||
__put_user(kbuf->f_namelen, &ubuf->f_namelen) ||
@@ -239,10 +262,34 @@ out:
static int put_compat_statfs64(struct compat_statfs64 __user *ubuf,
struct kstatfs *kbuf)
{
+ /*
+ * Large filesystems often have block counts that cannot be
represented in 32 bits.
+ * Rather than returning EFAULT to 32-bit apps in compatibility
mode, we fudge
+ * the blocksize upwards to keep the block count < 2^32
+ *
+ * For now the maximum 'fudge factor' is 16, i.e. the block size
will be <<4
+ * and the block count will be >>4.
+ */
+ int block_shift = 0;
+
if (sizeof ubuf->f_blocks == 4) {
if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
- 0xffffffff00000000ULL)
+ 0xfffffff000000000ULL)
return -EOVERFLOW;
+
+ if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
+ 0xfffffff800000000ULL)
+ block_shift = 4;
+ else if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
+ 0xfffffffc00000000ULL)
+ block_shift = 3;
+ else if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
+ 0xfffffffe00000000ULL)
+ block_shift = 2;
+ else if ((kbuf->f_blocks | kbuf->f_bfree | kbuf->f_bavail) &
+ 0xffffffff00000000ULL)
+ block_shift = 1;
+
/* f_files and f_ffree may be -1; it's okay
* to stuff that into 32 bits */
if (kbuf->f_files != 0xffffffffffffffffULL
@@ -254,10 +301,10 @@ static int put_compat_statfs64(struct co
}
if (!access_ok(VERIFY_WRITE, ubuf, sizeof(*ubuf)) ||
__put_user(kbuf->f_type, &ubuf->f_type) ||
- __put_user(kbuf->f_bsize, &ubuf->f_bsize) ||
- __put_user(kbuf->f_blocks, &ubuf->f_blocks) ||
- __put_user(kbuf->f_bfree, &ubuf->f_bfree) ||
- __put_user(kbuf->f_bavail, &ubuf->f_bavail) ||
+ __put_user(kbuf->f_bsize << block_shift, &ubuf->f_bsize) ||
+ __put_user(kbuf->f_blocks >> block_shift, &ubuf->f_blocks) ||
+ __put_user(kbuf->f_bfree >> block_shift, &ubuf->f_bfree) ||
+ __put_user(kbuf->f_bavail >> block_shift, &ubuf->f_bavail) ||
__put_user(kbuf->f_files, &ubuf->f_files) ||
__put_user(kbuf->f_ffree, &ubuf->f_ffree) ||
__put_user(kbuf->f_namelen, &ubuf->f_namelen) ||
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2006-09-29 01:55    [W:0.046 / U:0.868 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site