* coff/pe/xcoff and --extract-symbols
@ 2023-07-28 4:39 Alan Modra
0 siblings, 0 replies; only message in thread
From: Alan Modra @ 2023-07-28 4:39 UTC (permalink / raw)
To: binutils
This fixes failure of the "extract symbols" test for rs6000, where
--extract-symbols generates a non-zero sized .text. By the look of
coffcode.h the same problem might occur for coff/pe too, but doesn't
happen to trigger a test failure.
bfd/
* coffcode.h (coff_compute_section_file_positions): Don't
adjust size of !SEC_LOAD sections.
binutils/
* objcopy.c (setup_section): Clear SEC_LOAD for --extract-symbol.
diff --git a/bfd/coffcode.h b/bfd/coffcode.h
index f37d029ef75..6c8080c841e 100644
--- a/bfd/coffcode.h
+++ b/bfd/coffcode.h
@@ -3176,8 +3176,9 @@ coff_compute_section_file_positions (bfd * abfd)
created and not just improve performances with gdb.
*/
- if ((!strcmp (current->name, _TEXT)
- || !strcmp (current->name, _DATA))
+ if ((current->flags & SEC_LOAD) != 0
+ && (!strcmp (current->name, _TEXT)
+ || !strcmp (current->name, _DATA))
&& (previous == NULL || strcmp(previous->name, _TDATA)))
{
bfd_vma align = 4096;
@@ -3190,7 +3191,8 @@ coff_compute_section_file_positions (bfd * abfd)
sofar += align + vma_off - sofar_off;
}
#endif
- if (previous != NULL)
+ if (previous != NULL
+ && (previous->flags & SEC_LOAD) != 0)
previous->size += sofar - old_sofar;
}
diff --git a/binutils/objcopy.c b/binutils/objcopy.c
index beb655cafc2..b2b7ab6ab7e 100644
--- a/binutils/objcopy.c
+++ b/binutils/objcopy.c
@@ -4161,13 +4161,25 @@ setup_section (bfd *ibfd, sec_ptr isection, void *obfdarg)
flags = p->flags | (flags & (SEC_HAS_CONTENTS | SEC_RELOC));
flags = check_new_section_flags (flags, obfd, bfd_section_name (isection));
}
- else if (strip_symbols == STRIP_NONDEBUG
- && (flags & (SEC_ALLOC | SEC_GROUP)) != 0
- && !is_nondebug_keep_contents_section (ibfd, isection))
+ else
{
- flagword clr = SEC_HAS_CONTENTS | SEC_LOAD | SEC_GROUP;
-
- if (bfd_get_flavour (obfd) == bfd_target_elf_flavour)
+ flagword clr = 0;
+
+ /* For --extract-symbols where section sizes are zeroed, clear
+ SEC_LOAD to indicate to coff_compute_section_file_positions that
+ section sizes should not be adjusted for ALIGN_SECTIONS_IN_FILE.
+ We don't want to clear SEC_HAS_CONTENTS as that will result
+ in symbols being classified as 'B' by nm. */
+ if (extract_symbol)
+ clr = SEC_LOAD;
+ /* If only keeping debug sections then we'll be keeping section
+ sizes in headers but making the sections have no contents. */
+ else if (strip_symbols == STRIP_NONDEBUG
+ && (flags & (SEC_ALLOC | SEC_GROUP)) != 0
+ && !is_nondebug_keep_contents_section (ibfd, isection))
+ clr = SEC_HAS_CONTENTS | SEC_LOAD | SEC_GROUP;
+
+ if (clr && bfd_get_flavour (obfd) == bfd_target_elf_flavour)
{
/* PR 29532: Copy group sections intact as otherwise we end up with
empty groups. This prevents separate debug info files from
@@ -4175,7 +4187,7 @@ setup_section (bfd *ibfd, sec_ptr isection, void *obfdarg)
originally contained groups. */
if (flags & SEC_GROUP)
clr = SEC_LOAD;
- else
+ if ((clr & SEC_HAS_CONTENTS) != 0)
make_nobits = true;
/* Twiddle the input section flags so that it seems to
--
Alan Modra
Australia Development Lab, IBM
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2023-07-28 4:40 UTC | newest]
Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-07-28 4:39 coff/pe/xcoff and --extract-symbols Alan Modra
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).