On Wed, Mar 08, 2017 at 01:19:12PM -0500, Trevor Woerner wrote:
On Wed 2017-03-08 @ 12:20:47 PM, Tom Rini wrote:
On Wed, Mar 08, 2017 at 11:57:58AM -0500, Trevor Woerner wrote:
On Wed 2017-03-08 @ 10:35:19 AM, Tom Rini wrote:
perhaps we can do -N --no-omgic (which negates most of what -N does
Unfortunately "-N --no-omagic" doesn't succeed either (same problem):
arm-oe-linux-gnueabi-ld.bfd: u-boot: Not enough room for program headers, try linking with -N
OK. Lets dig a little harder. Do you have an older binutils around as well? It would be good to compare the objdump between the older version and newer version with -N passed in as well. Thanks!
No problem, OpenEmbedded FTW!! ;-) (it's quite easy to switch between them and build, using OE)
Would comparing the map files suffice? If I build u-boot twice (once with binutils-2.27 and once with binutils-2.28) and add the "-N" flag to the linking for both builds, comparing the map files I get:
$ diff -u 2.27-with-N/u-boot.map 2.28-with-N/u-boot.map --- 2.27-with-N/u-boot.map 2017-03-08 13:05:32.380107036 -0500 +++ 2.28-with-N/u-boot.map 2017-03-08 13:07:50.240330617 -0500 @@ -7084,6 +7084,7 @@
.data 0x0000000000043678 0x296c *(.data*)
- .data.rel.ro 0x0000000000043678 0x0 arch/arm/cpu/armv7/start.o .data.rk3288_otg_data 0x0000000000043678 0x2c arch/arm/mach-rockchip/built-in.o .data.rk3288_dmc_ops
@@ -7946,6 +7947,9 @@ *(.rel*) .rel.got 0x0000000000047cc0 0x0 arch/arm/cpu/armv7/start.o .rel.plt 0x0000000000047cc0 0x0 arch/arm/cpu/armv7/start.o
- .rel.bss 0x0000000000047cc0 0x0 arch/arm/cpu/armv7/start.o
- .rel.data.rel.ro
0x0000000000047cc0 0x0 arch/arm/cpu/armv7/start.oIf this isn't what you're looking for, please suggest what you'd like and I'll give it a try.
That's very close :) What does 2.27 without -N look like compared with 2.28 with -N ?