Re: [PATCH 2/2] lockfile: set lk->fd = -1 in 'rollback_lock_file()'

Re: [PATCH 2/2] lockfile: set lk->fd = -1 in 'rollback_lock_file()'

[Junio C Hamano]

Guanqun Lu <guanqun.lu@gmail.com> writes:

> Post-conditions when these functions return successfully:
>                     lk->fd == -1?    lk->filename[0] == '\0'?
> close_lock_file()       yes                 no
> commit_lock_file()      yes                 yes
> rollback_lock_file()    no*                 yes
>
> [*] This commit changes this 'no' in rollback_lock_file() to 'yes',
> which achieves more robust and unified interface.

Is there a broken caller this patch fixes?  IOW, is there a codepath that
can call rollback and then later notices lk->fd is not negative and uses
it without checking lk->filename[0] first?

> Signed-off-by: Guanqun Lu <guanqun.lu@gmail.com>
> ---
>  lockfile.c |    4 +++-
>  1 files changed, 3 insertions(+), 1 deletions(-)
>
> diff --git a/lockfile.c b/lockfile.c
> index 021c337..44e5253 100644
> --- a/lockfile.c
> +++ b/lockfile.c
> @@ -243,8 +243,10 @@ int commit_locked_index(struct lock_file *lk)
>  void rollback_lock_file(struct lock_file *lk)
>  {
>  	if (lk->filename[0]) {
> -		if (lk->fd >= 0)
> +		if (lk->fd >= 0) {
>  			close(lk->fd);
> +			lk->fd = -1;
> +		}
>  		unlink(lk->filename);
>  	}
>  	lk->filename[0] = 0;
> -- 
> 1.6.1.2.392.gb04d1

Re: [PATCH 1/2] code cleanup in sha1_file.c: xmalloc -> xcalloc

[Brandon Casey]

Guanqun Lu wrote:
> 'xmalloc' followed immediately by 'memset' is replaced
> with 'xcalloc', and a simple grep in this project seems
> to show that it's the only place.
> 
> Signed-off-by: Guanqun Lu <guanqun.lu@gmail.com>
> ---
>  sha1_file.c |    3 +--
>  1 files changed, 1 insertions(+), 2 deletions(-)
> 
> diff --git a/sha1_file.c b/sha1_file.c
> index 8868b80..93e5fc0 100644
> --- a/sha1_file.c
> +++ b/sha1_file.c
> @@ -858,8 +858,7 @@ unsigned char* use_pack(struct packed_git *p,
>  
>  static struct packed_git *alloc_packed_git(int extra)
>  {
> -	struct packed_git *p = xmalloc(sizeof(*p) + extra);
> -	memset(p, 0, sizeof(*p));
> +	struct packed_git *p = xcalloc(1, sizeof(*p) + extra);
>  	p->pack_fd = -1;
>  	return p;
>  }

I previously scanned through and did this.  I left this one as is
because the extra part is about as large as the sizeof(*p) part.  69 and
72 bytes respectively on 32-bit.  The extra part is always filled in
immediately by callers.  It's only called once for each pack so it's
not performance critical, so maybe your patch is worth it since it is
simpler?

-brandon

Re: [PATCH 1/2] code cleanup in sha1_file.c: xmalloc -> xcalloc

[Junio C Hamano]

Guanqun Lu <guanqun.lu@gmail.com> writes:

> 'xmalloc' followed immediately by 'memset' is replaced
> with 'xcalloc', and a simple grep in this project seems
> to show that it's the only place.

But isn't this memset() done only for the initial part of the allocated
area, not the whole thing?  You are not cleaning up but changing what it
does, if I am reading this code correctly.

> Signed-off-by: Guanqun Lu <guanqun.lu@gmail.com>
> ---
>  sha1_file.c |    3 +--
>  1 files changed, 1 insertions(+), 2 deletions(-)
>
> diff --git a/sha1_file.c b/sha1_file.c
> index 8868b80..93e5fc0 100644
> --- a/sha1_file.c
> +++ b/sha1_file.c
> @@ -858,8 +858,7 @@ unsigned char* use_pack(struct packed_git *p,
>  
>  static struct packed_git *alloc_packed_git(int extra)
>  {
> -	struct packed_git *p = xmalloc(sizeof(*p) + extra);
> -	memset(p, 0, sizeof(*p));
> +	struct packed_git *p = xcalloc(1, sizeof(*p) + extra);
>  	p->pack_fd = -1;
>  	return p;
>  }
> -- 
> 1.6.1.2.392.gb04d1

[PATCH 1/2] code cleanup in sha1_file.c: xmalloc -> xcalloc

[Guanqun Lu]

'xmalloc' followed immediately by 'memset' is replaced
with 'xcalloc', and a simple grep in this project seems
to show that it's the only place.

Signed-off-by: Guanqun Lu <guanqun.lu@gmail.com>
---
 sha1_file.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)

diff --git a/sha1_file.c b/sha1_file.c
index 8868b80..93e5fc0 100644
--- a/sha1_file.c
+++ b/sha1_file.c
@@ -858,8 +858,7 @@ unsigned char* use_pack(struct packed_git *p,
 
 static struct packed_git *alloc_packed_git(int extra)
 {
-	struct packed_git *p = xmalloc(sizeof(*p) + extra);
-	memset(p, 0, sizeof(*p));
+	struct packed_git *p = xcalloc(1, sizeof(*p) + extra);
 	p->pack_fd = -1;
 	return p;
 }
-- 
1.6.1.2.392.gb04d1

[PATCH 2/2] lockfile: set lk->fd = -1 in 'rollback_lock_file()'

[Guanqun Lu]

Post-conditions when these functions return successfully:
                    lk->fd == -1?    lk->filename[0] == '\0'?
close_lock_file()       yes                 no
commit_lock_file()      yes                 yes
rollback_lock_file()    no*                 yes

[*] This commit changes this 'no' in rollback_lock_file() to 'yes',
which achieves more robust and unified interface.

Signed-off-by: Guanqun Lu <guanqun.lu@gmail.com>
---
 lockfile.c |    4 +++-
 1 files changed, 3 insertions(+), 1 deletions(-)

diff --git a/lockfile.c b/lockfile.c
index 021c337..44e5253 100644
--- a/lockfile.c
+++ b/lockfile.c
@@ -243,8 +243,10 @@ int commit_locked_index(struct lock_file *lk)
 void rollback_lock_file(struct lock_file *lk)
 {
 	if (lk->filename[0]) {
-		if (lk->fd >= 0)
+		if (lk->fd >= 0) {
 			close(lk->fd);
+			lk->fd = -1;
+		}
 		unlink(lk->filename);
 	}
 	lk->filename[0] = 0;
-- 
1.6.1.2.392.gb04d1