threads / patch / 48088

patch, 2 partsread-cache: fix an -Wmaybe-uninitialized warning

Subject: [PATCH 2/2] read-cache: fix an -Wmaybe-uninitialized warning

## tl;dr

4 messages between Mar 19, 2018 and Mar 21, 2018. Diffs are folded; open one to read it.

replies: 3people: 2as markdown or json

Ramsay Jones· Mar 19, 2018, 17:56 UTC · lore

The function ce_write_entry() uses a 'self-initialised' variable construct, for the symbol 'saved_namelen', to suppress a gcc '-Wmaybe-uninitialized' warning, given that the warning is a false positive.

For the purposes of this discussion, the ce_write_entry() function has three code blocks of interest, that look like so:

        /* block #1 */
        if (ce->ce_flags & CE_STRIP_NAME) {
                saved_namelen = ce_namelen(ce);
                ce->ce_namelen = 0;
        }

        /* block #2 */
        /*
	 * several code blocks that contain, among others, calls
         * to copy_cache_entry_to_ondisk(ondisk, ce);
         */

        /* block #3 */
        if (ce->ce_flags & CE_STRIP_NAME) {
                ce->ce_namelen = saved_namelen;
                ce->ce_flags &= ~CE_STRIP_NAME;
        }

The warning implies that gcc thinks it is possible that the first block is not entered, the calls to copy_cache_entry_to_ondisk() could toggle the CE_STRIP_NAME flag on, thereby entering block #3 with saved_namelen unset. However, the copy_cache_entry_to_ondisk() function does not write to ce->ce_flags (it only reads). gcc could easily determine this, since that function is local to this file, but it obviously doesn't.

In order to suppress this warning, we make it clear to the reader (human and compiler), that block #3 will only be entered when the first block has been entered, by introducing a new 'stripped_name' boolean variable. We also take the opportunity to change the type of 'saved_namelen' to 'unsigned int' to match ce->ce_namelen.

Signed-off-by: Ramsay Jones <ramsay@ramsayjones.plus.com>
---
 read-cache.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

Show changes to read-cache.c +4 −2

diff --git a/read-cache.c b/read-cache.c
index 2eb81a66b..49607ddcd 100644
--- a/read-cache.c
+++ b/read-cache.c
@@ -2104,13 +2104,15 @@ static int ce_write_entry(git_SHA_CTX *c, int fd, struct cache_entry *ce,
 			  struct strbuf *previous_name, struct ondisk_cache_entry *ondisk)
 {
 	int size;
-	int saved_namelen = saved_namelen; /* compiler workaround */
 	int result;
+	unsigned int saved_namelen;
+	int stripped_name = 0;
 	static unsigned char padding[8] = { 0x00 };
 
 	if (ce->ce_flags & CE_STRIP_NAME) {
 		saved_namelen = ce_namelen(ce);
 		ce->ce_namelen = 0;
+		stripped_name = 1;
 	}
 
 	if (ce->ce_flags & CE_EXTENDED)
@@ -2150,7 +2152,7 @@ static int ce_write_entry(git_SHA_CTX *c, int fd, struct cache_entry *ce,
 		strbuf_splice(previous_name, common, to_remove,
 			      ce->name + common, ce_namelen(ce) - common);
 	}
-	if (ce->ce_flags & CE_STRIP_NAME) {
+	if (stripped_name) {
 		ce->ce_namelen = saved_namelen;
 		ce->ce_flags &= ~CE_STRIP_NAME;
 	}

-- 
2.16.0

Jeff King· Mar 20, 2018, 04:36 UTC · re: Ramsay Jones · lore

Re: [PATCH 2/2] read-cache: fix an -Wmaybe-uninitialized warning

On Mon, Mar 19, 2018 at 05:56:11PM +0000, Ramsay Jones wrote:

Show 28 quoted lines

> For the purposes of this discussion, the ce_write_entry() function has
> three code blocks of interest, that look like so:
> 
>         /* block #1 */
>         if (ce->ce_flags & CE_STRIP_NAME) {
>                 saved_namelen = ce_namelen(ce);
>                 ce->ce_namelen = 0;
>         }
> 
>         /* block #2 */
>         /*
> 	 * several code blocks that contain, among others, calls
>          * to copy_cache_entry_to_ondisk(ondisk, ce);
>          */
> 
>         /* block #3 */
>         if (ce->ce_flags & CE_STRIP_NAME) {
>                 ce->ce_namelen = saved_namelen;
>                 ce->ce_flags &= ~CE_STRIP_NAME;
>         }
> 
> The warning implies that gcc thinks it is possible that the first
> block is not entered, the calls to copy_cache_entry_to_ondisk()
> could toggle the CE_STRIP_NAME flag on, thereby entering block #3
> with saved_namelen unset. However, the copy_cache_entry_to_ondisk()
> function does not write to ce->ce_flags (it only reads). gcc could
> easily determine this, since that function is local to this file,
> but it obviously doesn't.

Weird. It seems like it would be pretty easy for it to know that we don't write the flags field at all. But I also don't see any other thing that would fool the compiler.

Show 5 quoted lines

> In order to suppress this warning, we make it clear to the reader
> (human and compiler), that block #3 will only be entered when the
> first block has been entered, by introducing a new 'stripped_name'
> boolean variable. We also take the opportunity to change the type
> of 'saved_namelen' to 'unsigned int' to match ce->ce_namelen.

These probably both ought to be size_t, but it makes sense to match ce_namelen for now.

Show 12 quoted lines

> diff --git a/read-cache.c b/read-cache.c
> index 2eb81a66b..49607ddcd 100644
> --- a/read-cache.c
> +++ b/read-cache.c
> @@ -2104,13 +2104,15 @@ static int ce_write_entry(git_SHA_CTX *c, int fd, struct cache_entry *ce,
>  			  struct strbuf *previous_name, struct ondisk_cache_entry *ondisk)
>  {
>  	int size;
> -	int saved_namelen = saved_namelen; /* compiler workaround */
>  	int result;
> +	unsigned int saved_namelen;
> +	int stripped_name = 0;

Maybe too clever, but I think you could just do:

  unsigned int saved_namelen = 0;
  ...
	saved_namelen = ce_namelen(ce);
  ...
  if (saved_namelen)
	ce->ce_namelen = saved_namelen;
  ce->ce_flags &= ~CE_STRIP_NAME;

the zero-length name case (if that's even legal) would work out the same.

That probably falls under the category of bikeshedding, though.

-Peff

Ramsay Jones· Mar 20, 2018, 22:52 UTC · re: Jeff King · lore

Re: [PATCH 2/2] read-cache: fix an -Wmaybe-uninitialized warning

On 20/03/18 04:36, Jeff King wrote:

> On Mon, Mar 19, 2018 at 05:56:11PM +0000, Ramsay Jones wrote:
>

[snip]

Show 25 quoted lines

>> diff --git a/read-cache.c b/read-cache.c
>> index 2eb81a66b..49607ddcd 100644
>> --- a/read-cache.c
>> +++ b/read-cache.c
>> @@ -2104,13 +2104,15 @@ static int ce_write_entry(git_SHA_CTX *c, int fd, struct cache_entry *ce,
>>  			  struct strbuf *previous_name, struct ondisk_cache_entry *ondisk)
>>  {
>>  	int size;
>> -	int saved_namelen = saved_namelen; /* compiler workaround */
>>  	int result;
>> +	unsigned int saved_namelen;
>> +	int stripped_name = 0;
> 
> Maybe too clever, but I think you could just do:
> 
>   unsigned int saved_namelen = 0;
>   ...
> 	saved_namelen = ce_namelen(ce);
>   ...
>   if (saved_namelen)
> 	ce->ce_namelen = saved_namelen;
>   ce->ce_flags &= ~CE_STRIP_NAME;
> 
> the zero-length name case (if that's even legal) would work out the
> same.

Yeah, that was one option that I looked at. The first option was to initialise saved_namelen to -1 (it was still an int) then the test became if (saved_namelen >= 0). However, that started me thinking about the zero-length case - should I assert if ((ce->ce_flags & CE_STRIP_NAME) && (ce_namelen(ce) == 0))? etc.

In the end, I decided that I wanted it to be 'drop dead' obvious what was going on! Hopefully, the result was just that. :-D

ATB, Ramsay Jones

Jeff King· Mar 21, 2018, 05:59 UTC · re: Ramsay Jones · lore

Re: [PATCH 2/2] read-cache: fix an -Wmaybe-uninitialized warning

On Tue, Mar 20, 2018 at 10:52:16PM +0000, Ramsay Jones wrote:

Show 21 quoted lines

> > Maybe too clever, but I think you could just do:
> > 
> >   unsigned int saved_namelen = 0;
> >   ...
> > 	saved_namelen = ce_namelen(ce);
> >   ...
> >   if (saved_namelen)
> > 	ce->ce_namelen = saved_namelen;
> >   ce->ce_flags &= ~CE_STRIP_NAME;
> > 
> > the zero-length name case (if that's even legal) would work out the
> > same.
> 
> Yeah, that was one option that I looked at. The first option
> was to initialise saved_namelen to -1 (it was still an int) then
> the test became if (saved_namelen >= 0). However, that started
> me thinking about the zero-length case - should I assert if
> ((ce->ce_flags & CE_STRIP_NAME) && (ce_namelen(ce) == 0))? etc.
> 
> In the end, I decided that I wanted it to be 'drop dead' obvious
> what was going on! Hopefully, the result was just that. :-D

Yeah, thinking on it more, simple and stupid is the right thing to do here. Thanks for a dose of sanity. :)

-Peff