From: Mike Galbraith <mikeg@wen-online.de>
To: linux-kernel <linux-kernel@vger.kernel.org>
Cc: Alexander Viro <viro@math.psu.edu>
Subject: [patch] Re: 2.5.7 rm -r in tmpfs problem
Date: Wed, 27 Mar 2002 11:24:46 +0100 (CET) [thread overview]
Message-ID: <Pine.LNX.4.10.10203271112410.639-100000@mikeg.wen-online.de> (raw)
In-Reply-To: <Pine.LNX.4.10.10203261736010.440-100000@mikeg.wen-online.de>
On Tue, 26 Mar 2002, Mike Galbraith wrote:
> On Sat, 23 Mar 2002, Alexander Viro wrote:
>
> > There are known problems with rm(1) on ramfs/tmpfs/etc. - the thing assumes
> > that offsets in directory remain stable after unlink(2), but locking changes
> > didn't affect.
>
> Ok, I finally found a wee bit of clue. I understand the assumption,
> and in virgin 2.5.4, that assumption is enforced... somefsckingwhere ;-)
I suppose that doesn't matter much though, since the assumption working
still doesn't make it right. (mentioned it because I can't figure out
why the behavior changed despite MUCH reading.. seemed odd/reportable)
Anyway, the hack below seems to work fine. Is there a better way?
-Mike
--- fs/libfs.c.org Sun Mar 24 13:20:39 2002
+++ fs/libfs.c Wed Mar 27 10:46:13 2002
@@ -29,6 +29,21 @@
return 0;
}
+static int seek_ok(struct dentry *dir, int offset)
+{
+ struct list_head *list;
+ int count = 2;
+
+ list = dir->d_subdirs.next;
+
+ while(list != &dir->d_subdirs) {
+ list = list->next;
+ count++;
+ }
+
+ return offset > count;
+}
+
/*
* Directory is locked and all positive dentries in it are safe, since
* for ramfs-type trees they can't go away without unlink() or rmdir(),
@@ -37,10 +52,9 @@
int dcache_readdir(struct file * filp, void * dirent, filldir_t filldir)
{
- int i;
+ int i = filp->f_pos;
struct dentry *dentry = filp->f_dentry;
- i = filp->f_pos;
switch (i) {
case 0:
if (filldir(dirent, ".", 1, i, dentry->d_inode->i_ino, DT_DIR) < 0)
@@ -57,11 +71,17 @@
default: {
struct list_head *list;
int j = i-2;
+ int abort;
spin_lock(&dcache_lock);
list = dentry->d_subdirs.next;
+ /*
+ * HACK: if we're attempting to seek _past_ the last
+ * entry, the directory is being modified. Abort seek.
+ */
+ abort = seek_ok(dentry, i);
- for (;;) {
+ for (;!abort;) {
if (list == &dentry->d_subdirs) {
spin_unlock(&dcache_lock);
return 0;
prev parent reply other threads:[~2002-03-27 10:05 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2002-03-22 11:52 2.5.7 rm -r in tmpfs problem Mike Galbraith
2002-03-23 15:36 ` [datapoint] " Mike Galbraith
2002-03-23 15:41 ` Alexander Viro
2002-03-23 16:04 ` Mike Galbraith
2002-03-26 16:55 ` Mike Galbraith
2002-03-27 10:24 ` Mike Galbraith [this message]
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=Pine.LNX.4.10.10203271112410.639-100000@mikeg.wen-online.de \
--to=mikeg@wen-online.de \
--cc=linux-kernel@vger.kernel.org \
--cc=viro@math.psu.edu \
/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;
as well as URLs for NNTP newsgroup(s).