bitbake-diffsigs: consider the situation where sigdata and siginfo files having the same hash values

For now, `bitbake-diffsigs -t <recipe> <task>' doesn't work. It always outputs nothing.

The problem is that bitbake-diffsigs are comparing sigdata and siginfo files that have the same hash value. This is not what we want. These two files are actually duplicates considering the purpose of bitbake-diffsigs. So we need to remove one of them so that bitbake-diffsigs could actually compare the correct signature files.

Signed-off-by Chen Qi <[email protected]>

---
  bitbake/bin/bitbake-diffsigs | 18 +++++++++++++++++-
  1 file changed, 17 insertions(+), 1 deletion(-)

bitbake/bin/bitbake-diffsigs +17 -1

index 08ae00d..196f0b7 100755
@@ -46,6 +46,12 @@ logger = logger_create('bitbake-diffsigs')
  def find_compare_task(bbhandler, pn, taskname):
      """ Find the most recent signature files for the specified PN/task and compare them """
 
+    def get_hashval(siginfo):
+        if siginfo.endswith('.siginfo'):
+            return siginfo.rpartition(':')[2].partition('_')[0]
+        else:
+            return siginfo.rpartition('.')[2]
+
      if not hasattr(bb.siggen, 'find_siginfo'):
          logger.error('Metadata does not support finding signature data files')
          sys.exit(1)

@@ -54,7 +60,7 @@ def find_compare_task(bbhandler, pn, taskname):
          taskname = 'do_%s' % taskname
 
      filedates = bb.siggen.find_siginfo(pn, taskname, None, bbhandler.config_data)
-    latestfiles = sorted(filedates.keys(), key=lambda f: filedates[f])[-2:]
+    latestfiles = sorted(filedates.keys(), key=lambda f: filedates[f])[-3:]
      if not latestfiles:
          logger.error('No sigdata files found matching %s %s' % (pn, taskname))
          sys.exit(1)

@@ -62,6 +68,16 @@ def find_compare_task(bbhandler, pn, taskname):
          logger.error('Only one matching sigdata file found for the specified task (%s %s)' % (pn, taskname))
          sys.exit(1)
      else:
+        # It's possible that latestfiles contain 3 elements and the first two have the same hash value.
+        # In this case, we delete the second element.
+        # The above case is actually the most common one. Because we may have sigdata file and siginfo
+        # file having the same hash value. Comparing such two files makes no sense.
+        if len(latestfiles) == 3:
+            hash0 = get_hashval(latestfiles[0])
+            hash1 = get_hashval(latestfiles[1])
+            if hash0 == hash1:
+                latestfiles.pop(1)
+
          # Define recursion callback
          def recursecb(key, hash1, hash2):
              hashes = [hash1, hash2]

ping

Any comment on this one?

Best Regards,
Chen Qi