do we still care about building with gcc-4.9?

[Alan Modra]

On Wed, Jul 15, 2026 at 03:10:05PM +0200, Jan Beulich wrote:
> On 15.07.2026 15:02, Alan Modra wrote:
> > gcc-4.9 does not like the parentheses or cast in AARCH64_FIELD when
> > used in an initialiser:
> > 
> > opcodes/aarch64-opc-2.c:29:3: error: initializer element is not constant
> >    {AARCH64_OPND_CLASS_INT_REG, "Rd", OPD_F_HAS_INSERTER | OPD_F_HAS_EXTRACTOR, {AARCH64_FIELD (0, 5)}, "an integer register"},
> >    ^
> > opcodes/aarch64-opc-2.c:29:3: error: (near initialization for ‘aarch64_operands[1].fields[0]’)
> > 
> 
> To answer your question - I'd much prefer if we did. One of my main test
> systems is at 4.8.5.

Here's another one.  I'll put this on the branch too.
gcc-4.9 doesn't like an empty initialiser.

binutils/readelf.c: In function ‘process_relocs’:
binutils/readelf.c:10229:5: error: missing initializer for field ‘sh_name’ of ‘Elf_Internal_Shdr’ [-Werror=missing-field-initializers]
     Elf_Internal_Shdr section = {};

	* readelf.c (process_relocs): Avoid gcc-4.9 compile error.

diff --git a/binutils/readelf.c b/binutils/readelf.c
index f18fbfbe8b9..a52fe2b8d63 100644
--- a/binutils/readelf.c
+++ b/binutils/readelf.c
@@ -10226,12 +10226,12 @@ process_relocs (Filedata * filedata)
 		     is passed to readelf.  */
 		  uint64_t num_reloc;
 		  uint64_t *relrs = NULL;
-		  Elf_Internal_Shdr section = {};
-		  section.sh_offset
-		    = filedata->dynamic_info[DT_RELR];
-		  section.sh_size = rel_size;
-		  section.sh_entsize = rel_entsz;
-		  section.sh_type = SHT_RELR;
+		  Elf_Internal_Shdr section = {
+		    .sh_type = SHT_RELR,
+		    .sh_offset = filedata->dynamic_info[DT_RELR],
+		    .sh_size = rel_size,
+		    .sh_entsize = rel_entsz
+		  };
 		  validate_section_info (&section, NULL, 0, filedata);
 		  num_reloc = count_relr_relocations (filedata,
 						      &section,


-- 
Alan Modra