Doesn't seem to work for me.
The -fno-stack-protector checking isn't working. After hacking the necessary makefiles to have -fno-stack-protector I get:
gcc -m32 -nostdlib -nostartfiles -static -o linuxbios -T ldscript.ld crt0.o /usr/bin/ld: section .id [00000000ffffef64 -> 00000000ffffef7f] overlaps section .rom [00000000ffff5e28 -> 00000000fffff00f] /usr/bin/ld: linuxbios: section .id lma 0xffffef64 overlaps previous sections /usr/bin/ld: linuxbios: section .reset lma 0xffffeff0 overlaps previous sections collect2: ld returned 1 exit status make[2]: *** [linuxbios] Error 1
Which IIRC means that my linuxbios image is too big. The build tutorial for this board seems to suggest that this is supposed to work. Is this a regression or is my Ubuntu gutsy toolchain to blame?
Doesn't seem to work for me.
The -fno-stack-protector checking isn't working. After hacking the necessary makefiles to have -fno-stack-protector I get:
gcc -m32 -nostdlib -nostartfiles -static -o linuxbios -T ldscript.ld crt0.o /usr/bin/ld: section .id [00000000ffffef64 -> 00000000ffffef7f] overlaps section .rom [00000000ffff5e28 -> 00000000fffff00f] /usr/bin/ld: linuxbios: section .id lma 0xffffef64 overlaps previous sections /usr/bin/ld: linuxbios: section .reset lma 0xffffeff0 overlaps previous sections collect2: ld returned 1 exit status make[2]: *** [linuxbios] Error 1
Which IIRC means that my linuxbios image is too big. The build tutorial for this board seems to suggest that this is supposed to work. Is this a regression or is my Ubuntu gutsy toolchain to blame?
If your payload isn't too big, try increasing the ROM_IMAGE_SIZE in Config.lb. You may still have room. You could also check the abuild Config file to see what size to set ROM_IMAGE_SIZE to.
Myles
On Sat, Dec 08, 2007 at 01:26:30PM -0500, Richard Smith wrote:
Doesn't seem to work for me.
The -fno-stack-protector checking isn't working.
Yep, known bug that needs fixing.
After hacking the necessary makefiles to have -fno-stack-protector I get:
Great, can you supply a patch?
gcc -m32 -nostdlib -nostartfiles -static -o linuxbios -T ldscript.ld crt0.o /usr/bin/ld: section .id [00000000ffffef64 -> 00000000ffffef7f] overlaps section .rom [00000000ffff5e28 -> 00000000fffff00f] /usr/bin/ld: linuxbios: section .id lma 0xffffef64 overlaps previous sections /usr/bin/ld: linuxbios: section .reset lma 0xffffeff0 overlaps previous sections collect2: ld returned 1 exit status make[2]: *** [linuxbios] Error 1
Which IIRC means that my linuxbios image is too big.
Yep. Did you do a make distclean && make ? I really have to do that after every change / failed build.
The build tutorial for this board seems to suggest that this is supposed to work. Is this a regression or is my Ubuntu gutsy toolchain to blame?
I build images just fine on a 32-bit Gnewsense (i.e. Dapper) box. Your rom chip is 1MB or bigger right (assuming you're doing LAB) - otherwise 512K is enough?
I've only tested file, etherboot and LAB payloads recently on this board. Kernel might work just fine too. Memtest probably needs a modified Config.lb.
Thanks, Ward.
On Sat, Dec 08, 2007 at 04:40:28PM -0500, Ward Vandewege wrote:
I build images just fine on a 32-bit Gnewsense (i.e. Dapper) box. Your rom chip is 1MB or bigger right (assuming you're doing LAB) - otherwise 512K is enough?
Ugh - that was clearly not relevant. The buildrom setup will make 512K images for filo/etherboot and 1MB images for kernel/LAB.
Toolchain problem?
Thanks, Ward.
Ward Vandewege wrote:
On Sat, Dec 08, 2007 at 01:26:30PM -0500, Richard Smith wrote:
Doesn't seem to work for me.
The -fno-stack-protector checking isn't working.
Yep, known bug that needs fixing.
After hacking the necessary makefiles to have -fno-stack-protector I get:
Great, can you supply a patch?
No. because its just a hack. Literally, I go edit the Makefiles and add a -fno-stack-protector. I have to add it in 3 places. The mkelfimage, and twice in linuxbios since the CAR code uses a different set of CFLAGS.
Jordan: I'll did into this a bit deeper, if want me to, but you will need to give me the 411 on what its supposed to be doing.
Yep. Did you do a make distclean && make ? I really have to do that after every change / failed build.
Yeah tried all that. I played with some settings as well. Busybox was still selected but if I de-select then I get:
# Ugh!!! Remember that libdl.a and libdl_pic.a are different. Since
# libdl is pretty small, and not likely to benefit from mklibs.py and # similar, lets just remove libdl_pic.a and avoid the issue rm -f /home/rsmith/projects/linuxbios/buildrom/buildrom-devel/staging/lib/libdl_pic.a make[1]: Leaving directory `/home/rsmith/projects/linuxbios/buildrom/buildrom-devel/work/uclibc/uClibc-0.9.28' cp: target `/home/rsmith/projects/linuxbios/buildrom/buildrom-devel/initrd-rootfs' is not a directory make: *** [/home/rsmith/projects/linuxbios/buildrom/buildrom-devel/deploy/initrd.uncompressed] Error 1
Is anyone else here on the list running gutsy?
On 10/12/07 13:26 -0500, Richard Smith wrote:
Ward Vandewege wrote:
On Sat, Dec 08, 2007 at 01:26:30PM -0500, Richard Smith wrote:
Doesn't seem to work for me.
The -fno-stack-protector checking isn't working.
Yep, known bug that needs fixing.
After hacking the necessary makefiles to have -fno-stack-protector I get:
Great, can you supply a patch?
No. because its just a hack. Literally, I go edit the Makefiles and add a -fno-stack-protector. I have to add it in 3 places. The mkelfimage, and twice in linuxbios since the CAR code uses a different set of CFLAGS.
Jordan: I'll did into this a bit deeper, if want me to, but you will need to give me the 411 on what its supposed to be doing.
I'll go in and fix it.
Jordan
richard, send me readelf -a of linuxbios. I may know what's going on.
ron
-
Jordan Crouse -
Myles Watson -
Richard Smith -
ron minnich -
Ward Vandewege