This is the mail archive of the binutils@sourceware.org mailing list for the binutils 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]

Delete redundant struct cie field


cie->output_sec is used to when merging CIEs to ensure that only CIEs
from the same output section are merged.  I noticed an assignment to
this field in _bfd_elf_parse_eh_frame, and thought "That's wrong,
output_section isn't set properly when _bfd_elf_parse_eh_frame is
called from gc-sections code".  It turns out that this assignment is
premature, and in fact a dead store.  find_merged_cie overwrites with
the correct value before the field is ever used.  On looking a little
more it becomes apparent that cie->cie_inf.u.cie.u.sec->output_section
holds the same value, so cie->output_sec is redundant.

	* elf-eh-frame.c (struct cie): Delete "output_sec" field.
	(cie_eq, cie_compute_hash): Use output_section from cie_inf instead.

diff --git a/bfd/elf-eh-frame.c b/bfd/elf-eh-frame.c
index 7783b08..d56440c 100644
--- a/bfd/elf-eh-frame.c
+++ b/bfd/elf-eh-frame.c
@@ -46,7 +46,6 @@ struct cie
     } sym;
     unsigned int reloc_index;
   } personality;
-  asection *output_sec;
   struct eh_cie_fde *cie_inf;
   unsigned char per_encoding;
   unsigned char lsda_encoding;
@@ -232,7 +231,8 @@ cie_eq (const void *e1, const void *e2)
       && c1->augmentation_size == c2->augmentation_size
       && memcmp (&c1->personality, &c2->personality,
 		 sizeof (c1->personality)) == 0
-      && c1->output_sec == c2->output_sec
+      && (c1->cie_inf->u.cie.u.sec->output_section
+	  == c2->cie_inf->u.cie.u.sec->output_section)
       && c1->per_encoding == c2->per_encoding
       && c1->lsda_encoding == c2->lsda_encoding
       && c1->fde_encoding == c2->fde_encoding
@@ -266,7 +266,7 @@ cie_compute_hash (struct cie *c)
   h = iterative_hash_object (c->ra_column, h);
   h = iterative_hash_object (c->augmentation_size, h);
   h = iterative_hash_object (c->personality, h);
-  h = iterative_hash_object (c->output_sec, h);
+  h = iterative_hash_object (c->cie_inf->u.cie.u.sec->output_section, h);
   h = iterative_hash_object (c->per_encoding, h);
   h = iterative_hash_object (c->lsda_encoding, h);
   h = iterative_hash_object (c->fde_encoding, h);
@@ -625,7 +625,6 @@ _bfd_elf_parse_eh_frame (bfd *abfd, struct bfd_link_info *info,
 
 	  cie->cie_inf = this_inf;
 	  cie->length = hdr_length;
-	  cie->output_sec = sec->output_section;
 	  start = buf;
 	  REQUIRE (read_byte (&buf, end, &cie->version));
 
@@ -1072,7 +1071,6 @@ find_merged_cie (bfd *abfd, struct bfd_link_info *info, asection *sec,
     }
 
   /* See if we can merge this CIE with an earlier one.  */
-  cie->output_sec = sec->output_section;
   cie_compute_hash (cie);
   if (hdr_info->cies == NULL)
     {

-- 
Alan Modra
Australia Development Lab, IBM


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