I have tried Andrew's fix. It worked. And I also tried to execute "rm -rf some_folder" in my yaffs2 partition, but I didn't see issues. Anybody knows what's that issue "rewind" fixed ? Thanks.
Regards,
Yang
Andrew,
Thanks for your info. I will try the fix. Btw , have you tested below fix with more cases?
I searched archived mails and found an old mailloop regarding the rewind change . Looks like that was for rm -rf issues of busybox ...
http://aleph1.co.uk/lurker/message/20051105.003401.0b8fec14.en.html
If we removed the "offset = 2;" , will it cause above issue again ? Or do we still need above workaround for the rm -rf issue for now ...As i understood , it was more like a application issue e.g busybox (i didn't check if bb has supported rewind now) , rather than the relatively low level file system issue.
Charles,
Appreciate your comments !
Regards,
Yang2008/11/5 Andrew McKay <amckay@iders.ca>
yang chen wrote:I just stopped the directory from rewinding if it was read or written. It would cause a loop where it would continuously rewind if there were more entries than could be buffered in a single read.
McKay,All
Any new update on this issue? Thanks. I just came across similar issue in which readdir(..) can't return NULL...causing endless loop...
Here's a patch. Not sure if it'll apply directly to your source, but should give you the idea of how I fixed it.
--- yaffs_fs.c 21 Dec 2007 00:24:37 -0000 1.1.8.1
+++ yaffs_fs.c 5 Sep 2008 19:36:29 -0000 1.1.8.1.24.1
@@ -960,7 +960,7 @@
readdir, rewind to after the 2 canned entries. */+ //offset = 2; //ARG!!! This is causing readdir to infinitely repeat directory listings. Don't Rewind!
if (f->f_version != inode->i_version) {
- offset = 2;Andrew McKay
f->f_pos = offset;
f->f_version = inode->i_version;
}
Iders Inc.