[PATCH] Replace memcpy with hashcpy when dealing hash copy globally

Signed-off-by: Sun He <redacted>
---
 Find the potential places with memcpy by the bash command:
   $ find . | xargs grep "memcpy.*\(.*20.*\)"

 Helped-by: Michael Haggerty[off-list ref]

 bundle.c            | 2 +-
 grep.c              | 2 +-
 pack-bitmap-write.c | 2 +-
 ppc/sha1.c          | 3 ++-
 reflog-walk.c       | 4 ++--
 refs.c              | 2 +-
 6 files changed, 8 insertions(+), 7 deletions(-)

diff --git a/bundle.c b/bundle.c
index e99065c..7809fbb 100644
--- a/bundle.c
+++ b/bundle.c

@@ -19,7 +19,7 @@ static void add_to_ref_list(const unsigned char *sha1, const char *name,
 		list->list = xrealloc(list->list,
 				list->alloc * sizeof(list->list[0]));
 	}
-	memcpy(list->list[list->nr].sha1, sha1, 20);
+	hashcpy(list->list[list->nr].sha1, sha1);
 	list->list[list->nr].name = xstrdup(name);
 	list->nr++;
 }

diff --git a/grep.c b/grep.c
index c668034..f5101f7 100644
--- a/grep.c
+++ b/grep.c

@@ -1650,7 +1650,7 @@ void grep_source_init(struct grep_source *gs, enum grep_source_type type,
 		break;
 	case GREP_SOURCE_SHA1:
 		gs->identifier = xmalloc(20);
-		memcpy(gs->identifier, identifier, 20);
+		hashcpy(gs->identifier, identifier);
 		break;
 	case GREP_SOURCE_BUF:
 		gs->identifier = NULL;

diff --git a/pack-bitmap-write.c b/pack-bitmap-write.c
index 1218bef..5f1791a 100644
--- a/pack-bitmap-write.c
+++ b/pack-bitmap-write.c

@@ -530,7 +530,7 @@ void bitmap_writer_finish(struct pack_idx_entry **index,
 	header.version = htons(default_version);
 	header.options = htons(flags | options);
 	header.entry_count = htonl(writer.selected_nr);
-	memcpy(header.checksum, writer.pack_checksum, 20);
+	hashcpy(header.checksum, writer.pack_checksum);
 
 	sha1write(f, &header, sizeof(header));
 	dump_bitmap(f, writer.commits);

diff --git a/ppc/sha1.c b/ppc/sha1.c
index ec6a192..8a87fea 100644
--- a/ppc/sha1.c
+++ b/ppc/sha1.c

@@ -9,6 +9,7 @@
 #include <stdio.h>
 #include <string.h>
 #include "sha1.h"
+#include "cache.h"
 
 extern void ppc_sha1_core(uint32_t *hash, const unsigned char *p,
 			  unsigned int nblocks);

@@ -67,6 +68,6 @@ int ppc_SHA1_Final(unsigned char *hash, ppc_SHA_CTX *c)
 		memset(&c->buf.b[cnt], 0, 56 - cnt);
 	c->buf.l[7] = c->len;
 	ppc_sha1_core(c->hash, c->buf.b, 1);
-	memcpy(hash, c->hash, 20);
+	hashcpy(hash, c->hash);
 	return 0;
 }

diff --git a/reflog-walk.c b/reflog-walk.c
index b2fbdb2..d490f7d 100644
--- a/reflog-walk.c
+++ b/reflog-walk.c

@@ -32,8 +32,8 @@ static int read_one_reflog(unsigned char *osha1, unsigned char *nsha1,
 			sizeof(struct reflog_info));
 	}
 	item = array->items + array->nr;
-	memcpy(item->osha1, osha1, 20);
-	memcpy(item->nsha1, nsha1, 20);
+	hashcpy(item->osha1, osha1);
+	hashcpy(item->nsha1, nsha1);
 	item->email = xstrdup(email);
 	item->timestamp = timestamp;
 	item->tz = tz;

diff --git a/refs.c b/refs.c
index 89228e2..f90b7ea 100644
--- a/refs.c
+++ b/refs.c

@@ -1222,7 +1222,7 @@ static int resolve_gitlink_packed_ref(struct ref_cache *refs,
 	if (ref == NULL)
 		return -1;
 
-	memcpy(sha1, ref->u.value.sha1, 20);
+	hashcpy(sha1, ref->u.value.sha1);
 	return 0;
 }
 

-- 
1.9.0.138.g2de3478.dirty

On Sat, Mar 1, 2014 at 8:07 AM, Sun He [off-list ref] wrote:

Signed-off-by: Sun He <redacted>
---
 Find the potential places with memcpy by the bash command:
   $ find . | xargs grep "memcpy.*\(.*20.*\)"

 Helped-by: Michael Haggerty[off-list ref]

You may want to put this Helped-by before "---" because it's supposed
to end up in the final commit. The patch looks straightforward,
except..

cache.h (actually git-compat-util.h that cache.h includes) messes
around with system headers by defining this and that macro. The
general rule is if cache.h or git-compat-util.h is included, it's the
first #include, and system includes will be always in
git-compat-util.h (grep '^#include' shows this). Maybe it's best to
leave this memcpy alone (and if you do, state so in the commit message
with the reason).
-- 
Duy

On Sat, Mar 1, 2014 at 10:58 AM, Duy Nguyen [off-list ref] wrote:

On Sat, Mar 1, 2014 at 8:07 AM, Sun He [off-list ref] wrote:

quoted

Signed-off-by: Sun He <redacted>
---
 Find the potential places with memcpy by the bash command:
   $ find . | xargs grep "memcpy.*\(.*20.*\)"

 Helped-by: Michael Haggerty[off-list ref]

You may want to put this Helped-by before "---" because it's supposed
to end up in the final commit.

To elaborate further: anything below the three-dash is ignored when
the patch is applied. So it doesn't show up in the commit at all. I
don't really know the reason for this - probably a patch (1) thing.

You could put the patch the Helped-by before your Signed-off-by
(sometimes referred to as S-o-b).

-- 
Cheers,
Ray Chuan

2014-03-01 10:58 GMT+08:00 Duy Nguyen [off-list ref]:

On Sat, Mar 1, 2014 at 8:07 AM, Sun He [off-list ref] wrote:

quoted

Signed-off-by: Sun He <redacted>
---
 Find the potential places with memcpy by the bash command:
   $ find . | xargs grep "memcpy.*\(.*20.*\)"

 Helped-by: Michael Haggerty[off-list ref]

You may want to put this Helped-by before "---" because it's supposed
to end up in the final commit. The patch looks straightforward,
except..

Yeah, got it.
Thanks.

quoted

diff --git a/ppc/sha1.c b/ppc/sha1.c
index ec6a192..8a87fea 100644
--- a/ppc/sha1.c
+++ b/ppc/sha1.c

@@ -9,6 +9,7 @@
 #include <stdio.h>
 #include <string.h>
 #include "sha1.h"
+#include "cache.h"

 extern void ppc_sha1_core(uint32_t *hash, const unsigned char *p,
                          unsigned int nblocks);

@@ -67,6 +68,6 @@ int ppc_SHA1_Final(unsigned char *hash, ppc_SHA_CTX *c)
                memset(&c->buf.b[cnt], 0, 56 - cnt);
        c->buf.l[7] = c->len;
        ppc_sha1_core(c->hash, c->buf.b, 1);
-       memcpy(hash, c->hash, 20);
+       hashcpy(hash, c->hash);
        return 0;
 }

cache.h (actually git-compat-util.h that cache.h includes) messes
around with system headers by defining this and that macro. The
general rule is if cache.h or git-compat-util.h is included, it's the
first #include, and system includes will be always in
git-compat-util.h (grep '^#include' shows this). Maybe it's best to
leave this memcpy alone (and if you do, state so in the commit message
with the reason).

Yap, after I parsed all the sourcecode

I have find out all the files that cache.h git-compat-util.h and
builtin.h are not the first #include

==> test-sigchain.c <==
#include "sigchain.h"
==> sigchain.c <==
#include "sigchain.h"

And I checked "sigchain.h", that it includes very little information.
It didn't import any potential errors. But I think it should be placed
after "cache.h" to match the consistence of the general rule.

--
Duy

2014-03-01 10:58 GMT+08:00 Duy Nguyen [off-list ref]:

On Sat, Mar 1, 2014 at 8:07 AM, Sun He [off-list ref] wrote:

quoted

Signed-off-by: Sun He <redacted>
---
 Find the potential places with memcpy by the bash command:
   $ find . | xargs grep "memcpy.*\(.*20.*\)"

 Helped-by: Michael Haggerty[off-list ref]

You may want to put this Helped-by before "---" because it's supposed
to end up in the final commit. The patch looks straightforward,
except..

Yeah, got it.
Thanks.

quoted

diff --git a/ppc/sha1.c b/ppc/sha1.c
index ec6a192..8a87fea 100644
--- a/ppc/sha1.c
+++ b/ppc/sha1.c

@@ -9,6 +9,7 @@
 #include <stdio.h>
 #include <string.h>
 #include "sha1.h"
+#include "cache.h"

 extern void ppc_sha1_core(uint32_t *hash, const unsigned char *p,
                          unsigned int nblocks);

@@ -67,6 +68,6 @@ int ppc_SHA1_Final(unsigned char *hash, ppc_SHA_CTX *c)
                memset(&c->buf.b[cnt], 0, 56 - cnt);
        c->buf.l[7] = c->len;
        ppc_sha1_core(c->hash, c->buf.b, 1);
-       memcpy(hash, c->hash, 20);
+       hashcpy(hash, c->hash);
        return 0;
 }

cache.h (actually git-compat-util.h that cache.h includes) messes
around with system headers by defining this and that macro. The
general rule is if cache.h or git-compat-util.h is included, it's the
first #include, and system includes will be always in
git-compat-util.h (grep '^#include' shows this). Maybe it's best to
leave this memcpy alone (and if you do, state so in the commit message
with the reason).

Yap, I should follow the general rule. My fault.
Thanks.

What's more,

I have find out all the files that cache.h git-compat-util.h and
builtin.h are not the first #include

==> test-sigchain.c <==
#include "sigchain.h"
==> sigchain.c <==
#include "sigchain.h"

And I checked "sigchain.h", that it includes very little information.
It didn't import any potential errors. But I think it should be placed
after "cache.h" to match the consistence of the general rule.

--
Duy

Got it.
Thanks.

2014-03-01 17:13 GMT+08:00 Tay Ray Chuan [off-list ref]:

On Sat, Mar 1, 2014 at 10:58 AM, Duy Nguyen [off-list ref] wrote:

quoted

On Sat, Mar 1, 2014 at 8:07 AM, Sun He [off-list ref] wrote:

quoted

Signed-off-by: Sun He <redacted>
---
 Find the potential places with memcpy by the bash command:
   $ find . | xargs grep "memcpy.*\(.*20.*\)"

 Helped-by: Michael Haggerty[off-list ref]

You may want to put this Helped-by before "---" because it's supposed
to end up in the final commit.

To elaborate further: anything below the three-dash is ignored when
the patch is applied. So it doesn't show up in the commit at all. I
don't really know the reason for this - probably a patch (1) thing.

You could put the patch the Helped-by before your Signed-off-by
(sometimes referred to as S-o-b).

--
Cheers,
Ray Chuan