This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Be prepared for DWARF4 .debug_frame (and .eh_frame) format

On 04/20/2010 02:04 AM, Jakub Jelinek wrote:
On Tue, Apr 20, 2010 at 01:19:55AM -0400, Jason Merrill wrote:
On 04/02/2010 05:32 AM, Jakub Jelinek wrote:
@@ -358,6 +358,12 @@ extract_cie_info (const struct dwarf_cie

    /* Immediately following the augmentation are the code and
       data alignment and return address column.  */
+  if (__builtin_expect (cie->version>= 4, 0))
+    {
+      if (p[0] != sizeof (void *) || p[1] != 0)
+	return NULL;
+      p += 2;
+    }

The comment needs an update.

+      if (aug[0] != 'z')
+	p += sizeof (void *);

This also needs a comment.

Like this?

I still don't understand what that's trying to skip. What is "g++ v2 pointer"? If there is any EH information it comes after the RA column, not immediately after the augmentation string.

-  /* Immediately following the augmentation are the code and
+  /* After the augmentation resp. pointer for "eh" augmentation
+     follows for CIE version>= 4 address size byte and
+     segment size byte.  */

Similarly, I don't see why we need to mention EH here.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]