[uClinux-dev] Help with relocation truncation problems on m5235 coldfire target

[Steve deRosier]

I hope someone here can help. I'm trying to port an older project
from ~5 year old version of uClinux to a current version. I've
managed to get everthing done, until I went to actually port the
business logic app. I'm getting the following types of errors when it
goes to link:
'relocation truncated to fit: R_68K_GOT16O'

Some searching and I roughly understand the problem, and the alleged
solution is to build some of the problem files with the -mxgot flag.
Fine. Then the following error:
'/usr/local/CodeSourcery/Sourcery_CodeBench_Lite_for_ColdFire_uClinux/m68k-uclinux/bin/elf2flt:
error: reloc type R_68K_GOT32O is not supported'

@#$%^#@!

Compile command line:
ucfront-gcc m68k-uclinux-gcc -mcpu=5235 -DCONFIG_COLDFIRE -O1 -g -pipe
-fno-common -fno-builtin -Wall -DEMBED -msep-data -Dlinux
-D__linux__ -Dunix -D__uClinux__ -Wall -Werror -Wno-unused
-Wpointer-arith -MD
-I/home/derosier/inthinc/projects/mcm820kernel/lib/libhessian/include
-I/home/derosier/inthinc/projects/mcm820kernel/user/iwi/src
-DCONFIG_5235 -DDIRECT_CALL -I./include -I/usr/include/libxml2
-I/home/derosier/inthinc/projects/mcm820kernel/staging/include
-I/home/derosier/inthinc/projects/mcm820kernel/user/iwi/src/common
-I/home/derosier/inthinc/projects/mcm820kernel/user/iwi/modules/sbs_module/include
-I/home/derosier/inthinc/projects/mcm820kernel/lib/libhessian/include
-c -o src/controller/queue.o src/controller/queue.c

When I keep the intermediates, here's the header at the top of the .s file:
| GNU C (Sourcery CodeBench Lite 2011.09-23) version 4.6.1 (m68k-uclinux)
| compiled by GNU C version 4.3.2, GMP version 4.3.2, MPFR version
3.0.1-p4, MPC version 0.9
| GGC heuristics: --param ggc-min-expand=100 --param ggc-min-heapsize=131072
| options passed: -fpreprocessed controller.i -mcpu=5235 -msep-data
| -auxbase-strip src/controller/controller.o -g -O1 -Wall -Werror
| -Wno-unused -Wpointer-arith -fno-common -fno-builtin -fverbose-asm
| options enabled: -fPIC -fauto-inc-dec -fbranch-count-reg
| -fcombine-stack-adjustments -fcompare-elim -fcprop-registers -fdefer-pop
| -fdelete-null-pointer-checks -fdwarf2-cfi-asm -fearly-inlining
| -feliminate-unused-debug-types -fforward-propagate -ffunction-cse
| -fgcse-lm -fguess-branch-probability -fident -fif-conversion
| -fif-conversion2 -finline -finline-functions-called-once -fipa-profile
| -fipa-pure-const -fipa-reference -fira-share-save-slots
| -fira-share-spill-slots -fivopts -fkeep-static-consts
| -fleading-underscore -fmath-errno -fmerge-constants -fmerge-debug-strings
| -fmove-loop-invariants -fpeephole -fprefetch-loop-arrays
| -freg-struct-return -fsched-critical-path-heuristic
| -fsched-dep-count-heuristic -fsched-group-heuristic -fsched-interblock
| -fsched-last-insn-heuristic -fsched-rank-heuristic -fsched-spec
| -fsched-spec-insn-heuristic -fsched-stalled-insns-dep -fshow-column
| -fsigned-zeros -fsplit-ivs-in-unroller -fsplit-wide-types
| -fstrict-volatile-bitfields -ftoplevel-reorder -ftrapping-math
| -ftree-bit-ccp -ftree-ccp -ftree-ch -ftree-copy-prop -ftree-copyrename
| -ftree-dce -ftree-dominator-opts -ftree-dse -ftree-forwprop -ftree-fre
| -ftree-loop-if-convert -ftree-loop-im -ftree-loop-ivcanon
| -ftree-loop-optimize -ftree-parallelize-loops= -ftree-phiprop -ftree-pta
| -ftree-reassoc -ftree-scev-cprop -ftree-sink -ftree-slp-vectorize
| -ftree-sra -ftree-ter -ftree-vect-loop-version -funit-at-a-time
| -fvar-tracking -fvar-tracking-assignments -fverbose-asm
| -fzero-initialized-in-bss -mdiv -msep-data


Link:

ucfront-gcc m68k-uclinux-gcc -mcpu=5235 -DCONFIG_COLDFIRE -Wl,-elf2flt
-Wl,-move-rodata -msep-data -Wl,-elf2flt -Wl,-move-rodata -msep-data
-L/home/derosier/inthinc/projects/mcm820kernel/user/iwi/libhessian -o
controller {...lots of .o...} libhessian/libhessian.a -lz -lm -lm

I suspect one of the fundamental problems is the code has some HUGE
functions (2500+ lines) with huge switch blocks, but I'm here mainly
to port not to muck with (ne refactor) the logic.

All help will be appreciated.

Thanks,
- Steve

[Larry Baker]

Steve,

I found this at http://gcc.gnu.org/onlinedocs/gcc-4.5.4/gcc/M680x0-Options.html. I don't know if gcc 4.3.2 has this option, but it's worth a try. Maybe you only have to use this in certain routines? Maybe not, if there is just one GOT for the entire program. Other google hits for "relocation truncated to fit: R_68K_GOT16O" said to remove XIP options or shared library, like -msep-data. I'm not sure that is possible on a ColdFire. I noticed you are using -O1. Did you try -Os? My ColdFire distribution uses -Os.

Post by Steve deRosier
-mxgot
-mno-xgot
When generating position-independent code for ColdFire, generate code that works if the GOT has more than 8192 entries. This code is larger and slower than code generated without this option. On M680x0 processors, this option is not needed; -fPIC suffices.
relocation truncated to fit: R_68K_GOT16O foobar
If this happens, you should recompile your code with -mxgot. It should then work with very large GOTs. However, code generated with -mxgot is less efficient, since it takes 4 instructions to fetch the value of a global symbol.
Note that some linkers, including newer versions of the GNU linker, can create multiple GOTs and sort GOT entries. If you have such a linker, you should only need to use -mxgotwhen compiling a single object file that accesses more than 8192 GOT entries. Very few do.
These options have no effect unless GCC is generating position-independent code.

Larry Baker
US Geological Survey
650-329-5608
baker at usgs.gov

Post by Steve deRosier
I hope someone here can help. I'm trying to port an older project
from ~5 year old version of uClinux to a current version. I've
managed to get everthing done, until I went to actually port the
business logic app. I'm getting the following types of errors when it
'relocation truncated to fit: R_68K_GOT16O'
Some searching and I roughly understand the problem, and the alleged
solution is to build some of the problem files with the -mxgot flag.
error: reloc type R_68K_GOT32O is not supported'
@#$%^#@!
ucfront-gcc m68k-uclinux-gcc -mcpu=5235 -DCONFIG_COLDFIRE -O1 -g -pipe
-fno-common -fno-builtin -Wall -DEMBED -msep-data -Dlinux
-D__linux__ -Dunix -D__uClinux__ -Wall -Werror -Wno-unused
-Wpointer-arith -MD
-I/home/derosier/inthinc/projects/mcm820kernel/lib/libhessian/include
-I/home/derosier/inthinc/projects/mcm820kernel/user/iwi/src
-DCONFIG_5235 -DDIRECT_CALL -I./include -I/usr/include/libxml2
-I/home/derosier/inthinc/projects/mcm820kernel/staging/include
-I/home/derosier/inthinc/projects/mcm820kernel/user/iwi/src/common
-I/home/derosier/inthinc/projects/mcm820kernel/user/iwi/modules/sbs_module/include
-I/home/derosier/inthinc/projects/mcm820kernel/lib/libhessian/include
-c -o src/controller/queue.o src/controller/queue.c
| GNU C (Sourcery CodeBench Lite 2011.09-23) version 4.6.1 (m68k-uclinux)
| compiled by GNU C version 4.3.2, GMP version 4.3.2, MPFR version
3.0.1-p4, MPC version 0.9
| GGC heuristics: --param ggc-min-expand=100 --param ggc-min-heapsize=131072
| options passed: -fpreprocessed controller.i -mcpu=5235 -msep-data
| -auxbase-strip src/controller/controller.o -g -O1 -Wall -Werror
| -Wno-unused -Wpointer-arith -fno-common -fno-builtin -fverbose-asm
| options enabled: -fPIC -fauto-inc-dec -fbranch-count-reg
| -fcombine-stack-adjustments -fcompare-elim -fcprop-registers -fdefer-pop
| -fdelete-null-pointer-checks -fdwarf2-cfi-asm -fearly-inlining
| -feliminate-unused-debug-types -fforward-propagate -ffunction-cse
| -fgcse-lm -fguess-branch-probability -fident -fif-conversion
| -fif-conversion2 -finline -finline-functions-called-once -fipa-profile
| -fipa-pure-const -fipa-reference -fira-share-save-slots
| -fira-share-spill-slots -fivopts -fkeep-static-consts
| -fleading-underscore -fmath-errno -fmerge-constants -fmerge-debug-strings
| -fmove-loop-invariants -fpeephole -fprefetch-loop-arrays
| -freg-struct-return -fsched-critical-path-heuristic
| -fsched-dep-count-heuristic -fsched-group-heuristic -fsched-interblock
| -fsched-last-insn-heuristic -fsched-rank-heuristic -fsched-spec
| -fsched-spec-insn-heuristic -fsched-stalled-insns-dep -fshow-column
| -fsigned-zeros -fsplit-ivs-in-unroller -fsplit-wide-types
| -fstrict-volatile-bitfields -ftoplevel-reorder -ftrapping-math
| -ftree-bit-ccp -ftree-ccp -ftree-ch -ftree-copy-prop -ftree-copyrename
| -ftree-dce -ftree-dominator-opts -ftree-dse -ftree-forwprop -ftree-fre
| -ftree-loop-if-convert -ftree-loop-im -ftree-loop-ivcanon
| -ftree-loop-optimize -ftree-parallelize-loops= -ftree-phiprop -ftree-pta
| -ftree-reassoc -ftree-scev-cprop -ftree-sink -ftree-slp-vectorize
| -ftree-sra -ftree-ter -ftree-vect-loop-version -funit-at-a-time
| -fvar-tracking -fvar-tracking-assignments -fverbose-asm
| -fzero-initialized-in-bss -mdiv -msep-data
ucfront-gcc m68k-uclinux-gcc -mcpu=5235 -DCONFIG_COLDFIRE -Wl,-elf2flt
-Wl,-move-rodata -msep-data -Wl,-elf2flt -Wl,-move-rodata -msep-data
-L/home/derosier/inthinc/projects/mcm820kernel/user/iwi/libhessian -o
controller {...lots of .o...} libhessian/libhessian.a -lz -lm -lm
I suspect one of the fundamental problems is the code has some HUGE
functions (2500+ lines) with huge switch blocks, but I'm here mainly
to port not to muck with (ne refactor) the logic.
All help will be appreciated.
Thanks,
- Steve
_______________________________________________
uClinux-dev mailing list
uClinux-dev at uclinux.org
http://mailman.uclinux.org/mailman/listinfo/uclinux-dev
This message was resent by uclinux-dev at uclinux.org
http://mailman.uclinux.org/mailman/options/uclinux-dev

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.uclinux.org/pipermail/uclinux-dev/attachments/20130208/44c6a2d5/attachment.html>

[Steve deRosier]

Hi Larry,

Post by Larry Baker
Steve,
I found this at
http://gcc.gnu.org/onlinedocs/gcc-4.5.4/gcc/M680x0-Options.html. I don't
know if gcc 4.3.2 has this option, but it's worth a try. Maybe you only
have to use this in certain routines? Maybe not, if there is just one GOT
R_68K_GOT16O" said to remove XIP options or shared library, like -msep-data.
I'm not sure that is possible on a ColdFire. I noticed you are using -O1.
Did you try -Os? My ColdFire distribution uses -Os.

Yes, my research came up with the -mxgot option also. Tried that and
came up with a problem where the elf2flt doesn't like the larger GOT
format. Aparently it's a problem my client has been fighting with for
a long time and I was able to define out a log statement and that
killed enough strings (5,000 log strings!) in the project to get it to
build. Basically I was told: "Get it to build and we'll deal with the
problem over here." Done.

-Os helped some also, but not with the GOT issue. But size-wise it
helped. Upgrading to Linux 3.2 from 2.4 (and the other apps) has
increased our total image size by nearly 50% and we're starting to
bump into size issues. I'm having to kill off some useful technician
and debugging utilities now.

Thanks for the help.

- Steve

[Philippe De Muyter]

Post by Steve deRosier
Hi Larry,

Post by Larry Baker
Steve,
I found this at
http://gcc.gnu.org/onlinedocs/gcc-4.5.4/gcc/M680x0-Options.html. I don't
know if gcc 4.3.2 has this option, but it's worth a try. Maybe you only
have to use this in certain routines? Maybe not, if there is just one GOT
R_68K_GOT16O" said to remove XIP options or shared library, like -msep-data.
I'm not sure that is possible on a ColdFire. I noticed you are using -O1.
Did you try -Os? My ColdFire distribution uses -Os.

Yes, my research came up with the -mxgot option also. Tried that and
came up with a problem where the elf2flt doesn't like the larger GOT
format. Aparently it's a problem my client has been fighting with for
a long time and I was able to define out a log statement and that
killed enough strings (5,000 log strings!) in the project to get it to
build. Basically I was told: "Get it to build and we'll deal with the
problem over here." Done.
-Os helped some also, but not with the GOT issue. But size-wise it
helped. Upgrading to Linux 3.2 from 2.4 (and the other apps) has
increased our total image size by nearly 50% and we're starting to
bump into size issues. I'm having to kill off some useful technician
and debugging utilities now.

Some years ago, I have written and submitted, but not pushed hard enough,
a patch to gcc for m68k permitting to generate an absolute text segment with
only a GOT table for the data and bss segments. Of course you must know
the load address at link-time, but that's doable if you do XIP.
This was combined with a patch to genromfs providing the load address
to the linker when generating the romfs image. This way I had an unique
absolute text segment for busybox with a small GOT table.

This must be still floating around

Hope that helps

Philippe