Commit graph

13 commits

Author SHA1 Message Date
Michael Sevakis
b4eec0dd42 Make INITDATA_ATTR work on everything that has INIT_ATTR enabled for code.
Change-Id: If9936bfbbd3bc3eb2a3e3e290701b8517eabfb13
2012-05-01 01:28:50 -04:00
Rafaël Carré
7ef13eed38 Fix path to PP crt0
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@31534 a1c6a512-1295-4272-9138-f99709370657
2012-01-03 04:44:27 +00:00
Boris Gjenero
ca9111ef64 Add KEEP() around vectors in linker scripts.
Vectors are needed by the CPU, but they don't need to be accessed by Rockbox.
Without the KEEP(), they can be removed when liking with --gc-sections, 
creating a broken binary without any warnings. This tells the linker to not
remove them. It should enable use of --gc-sections for all targets. When not
using --gc-sections, this does not change the binary.


git-svn-id: svn://svn.rockbox.org/rockbox/trunk@31351 a1c6a512-1295-4272-9138-f99709370657
2011-12-18 06:43:08 +00:00
Michael Sevakis
28dec004c0 Do some things to make -ffunction-sections work better.
* Add wildcards to various sections placements a la *(".text") => "*(.text*)"
* Remove hacky bits from those linker scripts (no problem encountered testing)
* Change section for asm functions from .<section> to .<section>.<function>
  so that -ffunction-sections works for those asm file too.


git-svn-id: svn://svn.rockbox.org/rockbox/trunk@31337 a1c6a512-1295-4272-9138-f99709370657
2011-12-17 01:43:32 +00:00
Boris Gjenero
e62dfa5225 FS#12397 : On targets which load .data directly into its final location and lack code for moving it, remove linker script trick which ignores section alignment and word-aligns the section instead.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@31041 a1c6a512-1295-4272-9138-f99709370657
2011-11-22 17:34:01 +00:00
Rafaël Carré
66f8fb52a9 */app.lds: remove STUBOFFSET
This is related to gdb, and gdb can only work on SH and ifp
This was mistakenly kept when app.lds was forked for each SoC
Side-effect: fix DEBUG builds when the rockbox binary is expected to
be loaded at the start of DRAM and there is no runtime relocation

git-svn-id: svn://svn.rockbox.org/rockbox/trunk@28124 a1c6a512-1295-4272-9138-f99709370657
2010-09-20 17:09:55 +00:00
Thomas Martitz
f8edc32589 FS#10756 - Free unused init code
Introduce a new .init section for initialisation code, so that it can be copied to an area which is later overwritten before calling. The stack/bss can then overwrite that code, effectively freeing the code size that the initialisation routines need. Gives a few kB ram usage back.
Only implemented for PP and as3525 so far. More targets could be added, as well as more functions.

git-svn-id: svn://svn.rockbox.org/rockbox/trunk@25013 a1c6a512-1295-4272-9138-f99709370657
2010-03-03 23:20:32 +00:00
Michael Sevakis
ec67912b63 Reclaim .iram areas in DRAM by overlapping their load addresses with the uninitialized data sections. I did what I could test out-- not any flash image linker scripts or other target processors. Move any .iram copies in crt0.S's to be the first operation even if not _strictly_ necessary to be emphatic (aka. 'beware').
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@20061 a1c6a512-1295-4272-9138-f99709370657
2009-02-20 02:33:40 +00:00
Jens Arnold
2bbacf89fe Make sure the linker considers crt0*.o before all files specified on the command line by using STARTUP() instead of INPUT() in the .lds. Using INPUT() makes the linker process crt0*.o after all files specified on the command line, making linking fail if there are librockbox.a members which are only referenced by crt0*.o. Remove the hackaround that was needed for building the gigabeat S bootloader before (aka UIE() problem).
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17289 a1c6a512-1295-4272-9138-f99709370657
2008-04-29 06:19:32 +00:00
Frank Gevaerts
7a2d9ddb08 move some data around to place qh_array at the start of iram. This saves up to about 2k iram on PP
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17194 a1c6a512-1295-4272-9138-f99709370657
2008-04-20 18:43:47 +00:00
Robert Kukla
feb3b588e3 explicitly align the ibss section by adjusting _iramend - fixes hang on m:robe (and potential problem on other arm targets)
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@17024 a1c6a512-1295-4272-9138-f99709370657
2008-04-07 18:24:23 +00:00
Michael Sevakis
05099149f1 Enable nocache sections using the linker. PP5022/4 must use SW_CORELOCK now with shared variables in DRAM (it seems swp(b) is at least partially broken on all PP or I'm doing something very wrong here :\). For core-shared data use SHAREDBSS/DATA_ATTR. NOCACHEBSS/DATA_ATTR is available whether or not single core is forced for static peripheral-DMA buffer allocation without use of the UNCACHED_ADDR macro in code and is likely useful on a non-PP target with a data cache (although not actually enabled in config.h and the .lds's in this commit).
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16981 a1c6a512-1295-4272-9138-f99709370657
2008-04-06 04:34:57 +00:00
Karl Kurbjun
ef62d6891a Split up app.lds to the respective target directories. The portalplayer devices replicate app.lds since their target tree doesn't follow a syntax typical to the newer arm targets - the portalplayers could be cleaned up further. boot.lds and plugin.lds still need to be cleaned up.
git-svn-id: svn://svn.rockbox.org/rockbox/trunk@16651 a1c6a512-1295-4272-9138-f99709370657
2008-03-13 03:48:23 +00:00