git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Linus Torvalds <torvalds@linux-foundation.org>
To: Junio C Hamano <gitster@pobox.com>,
	Git Mailing List <git@vger.kernel.org>
Subject: [PATCH 0/5] Be careful about lstat()-vs-readlink()
Date: Wed, 17 Dec 2008 10:42:02 -0800 (PST)	[thread overview]
Message-ID: <alpine.LFD.2.00.0812171034520.14014@localhost.localdomain> (raw)



This series of five patches makes us a lot more careful about readlink(), 
and in particular it avoids the code that depends on doing an lstat(), and 
then using st_size as the length of the link. That's what POSIX says 
_should_ happen, but Ramon Tayag reported git failing on NTFS under Linux 
because st_size doesn't match readlink() return value.

It doesn't seem to be unknown elsewhere either, since coreutils (through 
gnulib's areadlink_with_size) also has a big compatibility layer around 
readlink().

This series has been 'tested' by running it with the appended totally 
hacky patch that fakes out the lstat() st_size values for symlinks, so it 
has actually had some testing. 

The complete series does:

Linus Torvalds (5):
  Add generic 'strbuf_readlink()' helper function
  Make 'ce_compare_link()' use the new 'strbuf_readlink()'
  Make 'index_path()' use 'strbuf_readlink()'
  Make 'diff_populate_filespec()' use the new 'strbuf_readlink()'
  Make 'prepare_temp_file()' ignore st_size for symlinks

 builtin-apply.c |    6 ++----
 diff.c          |   25 +++++++++++--------------
 read-cache.c    |   22 ++++++++--------------
 sha1_file.c     |   14 +++++---------
 strbuf.c        |   28 ++++++++++++++++++++++++++++
 strbuf.h        |    1 +
 6 files changed, 55 insertions(+), 41 deletions(-)

and the hacky test-patch (which is obviously _not_ meant to be applied) is 
as follows...

		Linus

---
diff --git a/cache.h b/cache.h
index 231c06d..2d85fca 100644
--- a/cache.h
+++ b/cache.h
@@ -943,4 +943,7 @@ void overlay_tree_on_cache(const char *tree_name, const char *prefix);
 char *alias_lookup(const char *alias);
 int split_cmdline(char *cmdline, const char ***argv);
 
+extern int gitlstat(const char *, struct stat *);
+#define lstat gitlstat
+
 #endif /* CACHE_H */
diff --git a/read-cache.c b/read-cache.c
index b1475ff..defbb20 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -15,6 +15,16 @@
 #include "revision.h"
 #include "blob.h"
 
+#undef lstat
+int gitlstat(const char *path, struct stat *buf)
+{
+	int retval = lstat(path, buf);
+	if (!retval && S_ISLNK(buf->st_mode))
+		buf->st_size = 1;
+	return retval;
+}
+#define lstat gitlstat
+
 /* Index extensions.
  *
  * The first letter should be 'A'..'Z' for extensions that are not

             reply	other threads:[~2008-12-17 18:44 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-12-17 18:42 Linus Torvalds [this message]
2008-12-17 18:42 ` [PATCH 1/5] Add generic 'strbuf_readlink()' helper function Linus Torvalds
2008-12-17 18:43   ` [PATCH 2/5] Make 'ce_compare_link()' use the new 'strbuf_readlink()' Linus Torvalds
2008-12-17 18:43     ` [PATCH 3/5] Make 'index_path()' use 'strbuf_readlink()' Linus Torvalds
2008-12-17 18:44       ` [PATCH 4/5] Make 'diff_populate_filespec()' use the new 'strbuf_readlink()' Linus Torvalds
2008-12-17 18:45         ` [PATCH 5/5] Make 'prepare_temp_file()' ignore st_size for symlinks Linus Torvalds
2008-12-17 20:37           ` [PATCH 6/5] make_absolute_path(): check bounds when seeing an overlong symlink Junio C Hamano
2008-12-17 20:37           ` [PATCH 7/5] builtin-blame.c: use strbuf_readlink() Junio C Hamano
2008-12-17 20:37           ` [PATCH 8/5] combine-diff.c: " Junio C Hamano
2008-12-17 21:02             ` Linus Torvalds
2008-12-17 21:34               ` Junio C Hamano
2008-12-17 20:37         ` [PATCH 4/5] Make 'diff_populate_filespec()' use the new 'strbuf_readlink()' Junio C Hamano
2008-12-18 12:11         ` Mark Burton
2008-12-18 16:55           ` Linus Torvalds
2008-12-18 17:41             ` René Scharfe
2008-12-18 17:49               ` Linus Torvalds
2008-12-18 17:56                 ` Olivier Galibert
2008-12-18 16:56           ` René Scharfe
2008-12-18 17:28             ` René Scharfe
2008-12-19 22:10           ` [PATCH] diff.c: fix pointer type warning René Scharfe
2008-12-19 23:09             ` Junio C Hamano
2008-12-17 20:37       ` [PATCH 3/5] Make 'index_path()' use 'strbuf_readlink()' Junio C Hamano
2008-12-17 21:26   ` [PATCH 1/5] Add generic 'strbuf_readlink()' helper function Jay Soffian
2008-12-17 21:44     ` Linus Torvalds
2008-12-23 10:05   ` [PATCH] strbuf_readlink semantics update Pierre Habouzit
2008-12-23 10:21     ` Pierre Habouzit
2008-12-23 18:16       ` Linus Torvalds
2008-12-24 10:11         ` Pierre Habouzit
2008-12-24 15:20           ` René Scharfe
2008-12-25  7:23             ` Junio C Hamano
2009-01-04 12:21               ` Pierre Habouzit
2009-01-06 20:41                 ` [PATCH] strbuf: instate cleanup rule in case of non-memory errors René Scharfe
2009-01-07 21:19                   ` Junio C Hamano

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=alpine.LFD.2.00.0812171034520.14014@localhost.localdomain \
    --to=torvalds@linux-foundation.org \
    --cc=git@vger.kernel.org \
    --cc=gitster@pobox.com \
    /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).