public inbox for gcc-cvs@sourceware.org
help / color / mirror / Atom feed
* [gcc r11-11367] Darwin: Partial reversion of r14-3648 (Inits Section).
@ 2024-04-25 19:21 Iain D Sandoe
  0 siblings, 0 replies; only message in thread
From: Iain D Sandoe @ 2024-04-25 19:21 UTC (permalink / raw)
  To: gcc-cvs

https://gcc.gnu.org/g:960ca4144ebb61d125f28fab17b57cd3b09b49fd

commit r11-11367-g960ca4144ebb61d125f28fab17b57cd3b09b49fd
Author: Iain Sandoe <iain@sandoe.co.uk>
Date:   Sun Sep 10 14:48:42 2023 +0100

    Darwin: Partial reversion of r14-3648 (Inits Section).
    
    Although the Darwin ABI places both hot and cold partitions in the same
    section (the linker can partition by name), this does not work with the
    current dwarf2out implementation.
    
    Since we do see global initialization code getting hot/cold splits, this
    patch places the cold parts into text_cold, and keeps the hot part in
    the correct Init section per ABI.
    
    TODO: figure out a way to allow us to match the ABI fully.
    
    gcc/ChangeLog:
    
            * config/darwin.c (darwin_function_section): Place unlikely
            executed global init code into the standard cold section.
    
    Signed-off-by: Iain Sandoe <iain@sandoe.co.uk>
    (cherry picked from commit 5b33b364652866165431aef1810af1e890229c5e)

Diff:
---
 gcc/config/darwin.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/gcc/config/darwin.c b/gcc/config/darwin.c
index a3a7c509e36..db0e80889ab 100644
--- a/gcc/config/darwin.c
+++ b/gcc/config/darwin.c
@@ -3848,19 +3848,22 @@ darwin_function_section (tree decl, enum node_frequency freq,
   if (decl && DECL_SECTION_NAME (decl) != NULL)
     return get_named_section (decl, NULL, 0);
 
+  /* We always put unlikely executed stuff in the cold section; we have to put
+     this ahead of the global init section, since partitioning within a section
+     breaks some assumptions made in the DWARF handling.  */
+  if (freq == NODE_FREQUENCY_UNLIKELY_EXECUTED)
+    return (use_coal) ? darwin_sections[text_cold_coal_section]
+		      : darwin_sections[text_cold_section];
+
   /* Intercept functions in global init; these are placed in separate sections.
-     FIXME: there should be some neater way to do this.  */
+     FIXME: there should be some neater way to do this, FIXME we should be able
+     to partition within a section.  */
   if (DECL_NAME (decl)
       && (strncmp (IDENTIFIER_POINTER (DECL_NAME (decl)), "_GLOBAL__sub_I", 14) == 0
 	  || strncmp (IDENTIFIER_POINTER (DECL_NAME (decl)),
 			 "__static_initialization_and_destruction", 39) == 0))
     return  darwin_sections[static_init_section];
 
-  /* We always put unlikely executed stuff in the cold section.  */
-  if (freq == NODE_FREQUENCY_UNLIKELY_EXECUTED)
-    return (use_coal) ? darwin_sections[text_cold_coal_section]
-		      : darwin_sections[text_cold_section];
-
   /* If we have LTO *and* feedback information, then let LTO handle
      the function ordering, it makes a better job (for normal, hot,
      startup and exit - hence the bailout for cold above).  */

^ permalink raw reply	[flat|nested] only message in thread

only message in thread, other threads:[~2024-04-25 19:21 UTC | newest]

Thread overview: (only message) (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-04-25 19:21 [gcc r11-11367] Darwin: Partial reversion of r14-3648 (Inits Section) Iain D Sandoe

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).