[PATCH] use a hash of the lock token as the suffix for PUT/MOVE

[PATCH] use a hash of the lock token as the suffix for PUT/MOVE

[Tay Ray Chuan]

After 753bc91 ("Remove the requirement opaquelocktoken uri scheme"),
lock tokens are in the URI forms in which they are received from the
server, eg. 'opaquelocktoken:', 'urn:uuid:'.

However, "start_put" (and consequently "start_move"), which attempts to
create a unique temporary file using the UUID of the lock token,
inadvertently uses the lock token in its URI form. These file
operations on the server may not be successful (specifically, in
Windows), due to the colon ':' character from the URI form of the lock
token in the file path.

This patch uses a hash of the lock token instead, guaranteeing only
"safe" characters (a-f, 0-9) are used in the file path.

The token's hash is generated when the lock token is received from the
server in handle_new_lock_ctx, minimizing the number of times of
hashing.

Signed-off-by: Tay Ray Chuan <rctay89@gmail.com>
---

This patch is a result of the discussion on "[PATCH] use lock token in
non-URI form in start_put"; you can read it at
http://kerneltrap.org/mailarchive/git/2009/2/7/4922094.

The decision to use a hash of the token is so that one can avoid
handling the URI scheme of the lock token, which may be a deeply
nested URI, or (the more likely scenario) contain "unsafe" characters
for a file name, such as colons, slashes and spaces.

 http-push.c          |   11 ++++++++++-
 t/t5540-http-push.sh |    7 +++++++
 2 files changed, 17 insertions(+), 1 deletions(-)

diff --git a/http-push.c b/http-push.c
index eefd64c..0a252dd 100644
--- a/http-push.c
+++ b/http-push.c
@@ -153,6 +153,7 @@ struct remote_lock
 	char *url;
 	char *owner;
 	char *token;
+	char *token_sha1_hex;
 	time_t start_time;
 	long timeout;
 	int refreshing;
@@ -558,7 +559,7 @@ static void start_put(struct transfer_request *request)

 	append_remote_object_url(&buf, remote->url, hex, 0);
 	strbuf_addstr(&buf, "_");
-	strbuf_addstr(&buf, request->lock->token);
+	strbuf_addstr(&buf, request->lock->token_sha1_hex);
 	request->url = strbuf_detach(&buf, NULL);

 	slot = get_active_slot();
@@ -1130,6 +1131,8 @@ static void handle_lockprop_ctx(struct xml_ctx *ctx, int tag_closed)
 static void handle_new_lock_ctx(struct xml_ctx *ctx, int tag_closed)
 {
 	struct remote_lock *lock = (struct remote_lock *)ctx->userData;
+	git_SHA_CTX sha_ctx;
+	unsigned char lock_token_sha1[20];

 	if (tag_closed && ctx->cdata) {
 		if (!strcmp(ctx->name, DAV_ACTIVELOCK_OWNER)) {
@@ -1142,6 +1145,12 @@ static void handle_new_lock_ctx(struct xml_ctx *ctx, int tag_closed)
 		} else if (!strcmp(ctx->name, DAV_ACTIVELOCK_TOKEN)) {
 			lock->token = xmalloc(strlen(ctx->cdata) + 1);
 			strcpy(lock->token, ctx->cdata);
+
+			git_SHA1_Init(&sha_ctx);
+			git_SHA1_Update(&sha_ctx, lock->token, strlen(lock->token));
+			git_SHA1_Final(lock_token_sha1, &sha_ctx);
+
+			lock->token_sha1_hex = sha1_to_hex(lock_token_sha1);
 		}
 	}
 }
diff --git a/t/t5540-http-push.sh b/t/t5540-http-push.sh
index c236b5e..11b3432 100755
--- a/t/t5540-http-push.sh
+++ b/t/t5540-http-push.sh
@@ -94,6 +94,13 @@ test_expect_success 'MKCOL sends directory names with trailing slashes' '

 '

+test_expect_success 'PUT and MOVE sends object to URLs with SHA-1 hash suffix' '
+
+	grep -P "\"(?:PUT|MOVE) .+objects/[\da-z]{2}/[\da-z]{38}_[\da-z\-]{40} HTTP/[0-9.]+\" 20\d" \
+		< "$HTTPD_ROOT_PATH"/access.log
+
+'
+
 stop_httpd

 test_done
-- 
1.6.1.2.278.g9a9e.dirty

Re: [PATCH] use a hash of the lock token as the suffix for PUT/MOVE

[Junio C Hamano]

Tay Ray Chuan <rctay89@gmail.com> writes:

> After 753bc91 ("Remove the requirement opaquelocktoken uri scheme"),
> lock tokens are in the URI forms in which they are received from the
> server, eg. 'opaquelocktoken:', 'urn:uuid:'.
>
> However, "start_put" (and consequently "start_move"), which attempts to
> create a unique temporary file using the UUID of the lock token,
> inadvertently uses the lock token in its URI form. These file
> operations on the server may not be successful (specifically, in
> Windows), due to the colon ':' character from the URI form of the lock
> token in the file path.
>
> This patch uses a hash of the lock token instead, guaranteeing only
> "safe" characters (a-f, 0-9) are used in the file path.
>
> The token's hash is generated when the lock token is received from the
> server in handle_new_lock_ctx, minimizing the number of times of
> hashing.
>
> Signed-off-by: Tay Ray Chuan <rctay89@gmail.com>

Thanks, very clearly written.

> diff --git a/http-push.c b/http-push.c
> index eefd64c..0a252dd 100644
> --- a/http-push.c
> +++ b/http-push.c
> @@ -153,6 +153,7 @@ struct remote_lock
>  	char *url;
>  	char *owner;
>  	char *token;
> +	char *token_sha1_hex;

At this point, this new field is only used as a unique suffix, and
"sha1_hex" is a implementation detail of the mechanism to guarantee the
uniqueness.  Naming things for what they are is preferred over naming
things for how they are crafted.  Hence:

	char tmpfile_suffix[41];

would be a better definition here.

> @@ -558,7 +559,7 @@ static void start_put(struct transfer_request *request)
>
>  	append_remote_object_url(&buf, remote->url, hex, 0);
>  	strbuf_addstr(&buf, "_");
> -	strbuf_addstr(&buf, request->lock->token);
> +	strbuf_addstr(&buf, request->lock->token_sha1_hex, 41);

And replace these two strbuf_addstr() with:

	strbuf_add(&buf, request->lock->tmpfile_suffix, 41);

> @@ -1130,6 +1131,8 @@ static void handle_lockprop_ctx(struct xml_ctx *ctx, int tag_closed)
>  static void handle_new_lock_ctx(struct xml_ctx *ctx, int tag_closed)
>  {
>  	struct remote_lock *lock = (struct remote_lock *)ctx->userData;
> +	git_SHA_CTX sha_ctx;
> +	unsigned char lock_token_sha1[20];
>
>  	if (tag_closed && ctx->cdata) {
>  		if (!strcmp(ctx->name, DAV_ACTIVELOCK_OWNER)) {
> @@ -1142,6 +1145,12 @@ static void handle_new_lock_ctx(struct xml_ctx *ctx, int tag_closed)
>  		} else if (!strcmp(ctx->name, DAV_ACTIVELOCK_TOKEN)) {
>  			lock->token = xmalloc(strlen(ctx->cdata) + 1);
>  			strcpy(lock->token, ctx->cdata);
> +
> +			git_SHA1_Init(&sha_ctx);
> +			git_SHA1_Update(&sha_ctx, lock->token, strlen(lock->token));
> +			git_SHA1_Final(lock_token_sha1, &sha_ctx);
> +
> +			lock->token_sha1_hex = sha1_to_hex(lock_token_sha1);

The last one is wrong because string returned by sha1_to_hex() is
volatile.

	lock->tmpfile_suffix[0] = '_';
        memcpy(lock->tmpfile_suffix + 1, sha1_to_hex(sha1));

Other than that, I think this is a good patch.

Re: [PATCH] use a hash of the lock token as the suffix for PUT/MOVE

[Tay Ray Chuan]

After 753bc91 ("Remove the requirement opaquelocktoken uri scheme"),
lock tokens are in the URI forms in which they are received from the
server, eg. 'opaquelocktoken:', 'urn:uuid:'.

However, "start_put" (and consequently "start_move"), which attempts to
create a unique temporary file using the UUID of the lock token,
inadvertently uses the lock token in its URI form. These file
operations on the server may not be successful (specifically, in
Windows), due to the colon ':' character from the URI form of the lock
token in the file path.

This patch uses a hash of the lock token instead, guaranteeing only
"safe" characters (a-f, 0-9) are used in the file path.

The token's hash is generated when the lock token is received from the
server in handle_new_lock_ctx, minimizing the number of times of
hashing.

Signed-off-by: Tay Ray Chuan <rctay89@gmail.com>
---

This patch is a result of the discussion on "[PATCH] use lock token in
non-URI form in start_put"; you can read it at
http://kerneltrap.org/mailarchive/git/2009/2/7/4922094.

The decision to use a hash of the token is so that one can avoid
handling the URI scheme of the lock token, which may be a deeply
nested URI, or (the more likely scenario) contain "unsafe" characters
for a file name, such as colons, slashes and spaces.

* use tmpfile_suffix instead of token_sha1_hex in remote_lock, as
suggested by Junio, and absorb '_' as a result
* memcpy string from sha1_to_hex to tmpfile_suffix

 http-push.c          |   13 +++++++++++--
 t/t5540-http-push.sh |    7 +++++++
 2 files changed, 18 insertions(+), 2 deletions(-)

diff --git a/http-push.c b/http-push.c
index eefd64c..edbff58 100644
--- a/http-push.c
+++ b/http-push.c
@@ -153,6 +153,7 @@ struct remote_lock
 	char *url;
 	char *owner;
 	char *token;
+	char tmpfile_suffix[41];
 	time_t start_time;
 	long timeout;
 	int refreshing;
@@ -557,8 +558,7 @@ static void start_put(struct transfer_request *request)
 	request->dest = strbuf_detach(&buf, NULL);

 	append_remote_object_url(&buf, remote->url, hex, 0);
-	strbuf_addstr(&buf, "_");
-	strbuf_addstr(&buf, request->lock->token);
+	strbuf_add(&buf, request->lock->tmpfile_suffix, 41);
 	request->url = strbuf_detach(&buf, NULL);

 	slot = get_active_slot();
@@ -1130,6 +1130,8 @@ static void handle_lockprop_ctx(struct xml_ctx *ctx, int tag_closed)
 static void handle_new_lock_ctx(struct xml_ctx *ctx, int tag_closed)
 {
 	struct remote_lock *lock = (struct remote_lock *)ctx->userData;
+	git_SHA_CTX sha_ctx;
+	unsigned char lock_token_sha1[20];

 	if (tag_closed && ctx->cdata) {
 		if (!strcmp(ctx->name, DAV_ACTIVELOCK_OWNER)) {
@@ -1142,6 +1144,13 @@ static void handle_new_lock_ctx(struct xml_ctx *ctx, int tag_closed)
 		} else if (!strcmp(ctx->name, DAV_ACTIVELOCK_TOKEN)) {
 			lock->token = xmalloc(strlen(ctx->cdata) + 1);
 			strcpy(lock->token, ctx->cdata);
+
+			git_SHA1_Init(&sha_ctx);
+			git_SHA1_Update(&sha_ctx, lock->token, strlen(lock->token));
+			git_SHA1_Final(lock_token_sha1, &sha_ctx);
+
+			lock->tmpfile_suffix[0] = '_';
+			memcpy(lock->tmpfile_suffix + 1, sha1_to_hex(lock_token_sha1), 40);
 		}
 	}
 }
diff --git a/t/t5540-http-push.sh b/t/t5540-http-push.sh
index c236b5e..11b3432 100755
--- a/t/t5540-http-push.sh
+++ b/t/t5540-http-push.sh
@@ -94,6 +94,13 @@ test_expect_success 'MKCOL sends directory names with trailing slashes' '

 '

+test_expect_success 'PUT and MOVE sends object to URLs with SHA-1 hash suffix' '
+
+	grep -P "\"(?:PUT|MOVE) .+objects/[\da-z]{2}/[\da-z]{38}_[\da-z\-]{40} HTTP/[0-9.]+\" 20\d" \
+		< "$HTTPD_ROOT_PATH"/access.log
+
+'
+
 stop_httpd

 test_done
-- 
1.6.1.2.278.g9a9e.dirty

Re: [PATCH] use a hash of the lock token as the suffix for PUT/MOVE

[Tay Ray Chuan]

Hi,

On Sat, Feb 14, 2009 at 4:34 PM, Junio C Hamano <gitster@pobox.com> wrote:
>
> Other than that, I think this is a good patch.
>

thanks for the encouraging words, I've sent in the modified patch already.

-- 
Cheers,
Ray Chuan