Refactor SHA*_file functions into a separate file. Generate them using a C macro.

Refactor SHA*_file functions into a separate file. Generate them using a C macro.

[Randall Spangler, 2010-01-29 00:04:58]

Refactoring SHA* functions into a separate file good.  Big macro scary.

If you really don't like the duplicated code, you can achieve the same thing
with an array of structs, something like:

struct sha_flavor {
  (void *init_func)();
  (void *update_func)();
  (uint8_t *final_func)();
  int blocksize;
  int digestsize;
}

Then you can have

sha_flavor SHA1_flavors[] = {
  {SHA1_init, SHA1_update, SHA1, final, SHA1_BLOCK_SIZE, SHA1_DIGEST_SIZE};
  // similar for other digest types
};

Then calculate_digest() can use the algorithm param as an index into the above
array.  The individual SHA*_file() functions can just be macros which call
calculate_digest().

http://codereview.chromium.org/558025/diff/1/2
File src/platform/vboot_reference/tests/Makefile (right):

http://codereview.chromium.org/558025/diff/1/2#newcode20
src/platform/vboot_reference/tests/Makefile:20: rm -f $(OBJS) sha_tests
verify_data digest_utility.o
Since $SRCS contains digest_utility.c, shouldn't need to specifically list
digest_utility.o here.

http://codereview.chromium.org/558025/diff/1/3
File src/platform/vboot_reference/tests/digest_utility.c (right):

http://codereview.chromium.org/558025/diff/1/3#newcode19
src/platform/vboot_reference/tests/digest_utility.c:19: #define
CreateDigestHelper(__algo__)                                    \
Ok, I'll admit, it's a clever use of the ## operator.  

It's also really hard to read, and will be painful to debug, since you can't set
breakpoints inside this code.

http://codereview.chromium.org/558025/diff/1/4
File src/platform/vboot_reference/tests/digest_utility.h (right):

http://codereview.chromium.org/558025/diff/1/4#newcode13
src/platform/vboot_reference/tests/digest_utility.h:13: /* Returns the SHA-1
digest of [input_file].
Of the data in the input_file?  (or similar rewording)

(I had to look at the implementation to determine whether the input file was the
filename or the file data itself.)

[gauravsh, 2010-01-29 00:52:21]

As we discussed, we can't use a structure with function pointers since the type
of arguments differ to the various SHA*_init/final/update functions. However, in
the future, we can refactor the SHA code to use single SHA_CTX structure.

http://codereview.chromium.org/558025/diff/1/2
File src/platform/vboot_reference/tests/Makefile (right):

http://codereview.chromium.org/558025/diff/1/2#newcode20
src/platform/vboot_reference/tests/Makefile:20: rm -f $(OBJS) sha_tests
verify_data digest_utility.o
On 2010/01/29 00:04:58, Randall Spangler wrote:
> Since $SRCS contains digest_utility.c, shouldn't need to specifically list
> digest_utility.o here.

True. Fixed.

http://codereview.chromium.org/558025/diff/1/3
File src/platform/vboot_reference/tests/digest_utility.c (right):

http://codereview.chromium.org/558025/diff/1/3#newcode19
src/platform/vboot_reference/tests/digest_utility.c:19: #define
CreateDigestHelper(__algo__)                                    \
On 2010/01/29 00:04:58, Randall Spangler wrote:
> Ok, I'll admit, it's a clever use of the ## operator.  
> 
> It's also really hard to read, and will be painful to debug, since you can't
set
> breakpoints inside this code.

Heh, yes. Switched back to the non-macro (duplicated) version.

http://codereview.chromium.org/558025/diff/1/4
File src/platform/vboot_reference/tests/digest_utility.h (right):

http://codereview.chromium.org/558025/diff/1/4#newcode13
src/platform/vboot_reference/tests/digest_utility.h:13: /* Returns the SHA-1
digest of [input_file].
On 2010/01/29 00:04:58, Randall Spangler wrote:
> Of the data in the input_file?  (or similar rewording)
> 
> (I had to look at the implementation to determine whether the input file was
the
> filename or the file data itself.)

Done.

[Randall Spangler, 2010-01-29 00:54:05]

lgtm