linux-kernel - Re: [PATCH] ROMFS 0 byte file read error

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

Date:	Wed, 30 Jul 2008 11:06:29 -0700 (PDT)
From:	Linus Torvalds <torvalds@...ux-foundation.org>
To:	Chris Fester <cfester@....com>
cc:	lkml <linux-kernel@...r.kernel.org>,
	Alexander Viro <viro@...iv.linux.org.uk>,
	Matt Waddel <matt.waddel@...escale.com>,
	Greg Ungerer <gerg@...pgear.com>
Subject: Re: [PATCH] ROMFS 0 byte file read error



On Wed, 30 Jul 2008, Chris Fester wrote:
>
> I've verified that the git tree at:
> git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
> 
> also has the zero byte file problem for romfs.  This patch
> fixes the problem.  Avoids calling romfs_copyfrom when in 
> the 0 size case.

Hmm. Who calls 'readpage()' with an offset past the end of the file 
anyway? This _really_ shouldn't matter.

But regardless, isn't the bug that 'romfs_readpage()' sets an error bit 
by default - ie even if there was no actual IO error? The thing is also 
very confused about types: if it really wants to be safe in "loff_t, then 
it had bettr not do the "min_t()" in just "unsigned long".

So this whole routine really seems to be much more broken than your patch 
implies, and your patch just works around some brokenness.

Of course, I think it's all 32-bit and some of these problems cannot 
actually happen, but wouldn't it be more obvious to rewrite it to have a 
separate error return ("result") and a variable saying how much of the 
page was filled ("filled")?

Something like this. UNTESTED. Pls verify and send back if this is ok. The 
patch is certainly bigger, but I think that the end result is more 
obvious.

		Linus

---
 fs/romfs/inode.c |   37 +++++++++++++++++++++++--------------
 1 files changed, 23 insertions(+), 14 deletions(-)

diff --git a/fs/romfs/inode.c b/fs/romfs/inode.c
index 8e51a2a..60d2f82 100644
--- a/fs/romfs/inode.c
+++ b/fs/romfs/inode.c
@@ -418,7 +418,8 @@ static int
 romfs_readpage(struct file *file, struct page * page)
 {
 	struct inode *inode = page->mapping->host;
-	loff_t offset, avail, readlen;
+	loff_t offset, size;
+	unsigned long filled;
 	void *buf;
 	int result = -EIO;
 
@@ -430,21 +431,29 @@ romfs_readpage(struct file *file, struct page * page)
 
 	/* 32 bit warning -- but not for us :) */
 	offset = page_offset(page);
-	if (offset < i_size_read(inode)) {
-		avail = inode->i_size-offset;
-		readlen = min_t(unsigned long, avail, PAGE_SIZE);
-		if (romfs_copyfrom(inode, buf, ROMFS_I(inode)->i_dataoffset+offset, readlen) == readlen) {
-			if (readlen < PAGE_SIZE) {
-				memset(buf + readlen,0,PAGE_SIZE-readlen);
-			}
-			SetPageUptodate(page);
-			result = 0;
+	size = i_size_read(inode);
+	filled = 0;
+	result = 0;
+	if (offset < size) {
+		unsigned long readlen;
+
+		size -= offset;
+		readlen = size > PAGE_SIZE ? PAGE_SIZE : size;
+
+		filled = romfs_copyfrom(inode, buf, ROMFS_I(inode)->i_dataoffset+offset, readlen);
+
+		if (filled != readlen) {
+			SetPageError(page);
+			filled = 0;
+			result = -EIO;
 		}
 	}
-	if (result) {
-		memset(buf, 0, PAGE_SIZE);
-		SetPageError(page);
-	}
+
+	if (filled < PAGE_SIZE)
+		memset(buf + filled, 0, PAGE_SIZE-filled);
+
+	if (!result)
+		SetPageUptodate(page);
 	flush_dcache_page(page);
 
 	unlock_page(page);
--
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/