[Bug ld/13812] Suboptimal default linker script cause "internal error: out of range error"
http://sourceware.org/bugzilla/show_bug.cgi?id=13812 Senthil Kumar Selvaraj changed: What|Removed |Added CC||senthil.thecoder at gmail ||dot com --- Comment #4 from Senthil Kumar Selvaraj 2013-03-12 17:53:02 UTC --- Actually, the linker relaxation code already considers cases where deletion of the ret in call/ret or jmp/ret might be unsafe - just that it missed this specific case. The case being that GCC puts the array containing the addresses of the labels in a separate section (.rodata). The safe ret deletion logic assumes that only relocations in the local section (containing the ret) could potentially have symbols whose values evaluate to the address containing the ret, but that is not true in this case. Looping over all sections in the BFD, and repeating the check for each section fixes the problem (see attached patch). -- Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/13812] Suboptimal default linker script cause "internal error: out of range error"
http://sourceware.org/bugzilla/show_bug.cgi?id=13812 Senthil Kumar Selvaraj changed: What|Removed |Added Status|ASSIGNED|NEW --- Comment #6 from Senthil Kumar Selvaraj 2013-03-12 17:57:33 UTC --- Oops - Disregard my previous comments - they were meant for bug #12494 -- Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/13812] Suboptimal default linker script cause "internal error: out of range error"
http://sourceware.org/bugzilla/show_bug.cgi?id=13812 Senthil Kumar Selvaraj changed: What|Removed |Added Status|NEW |ASSIGNED AssignedTo|unassigned at sourceware|senthil.thecoder at gmail |dot org |dot com --- Comment #5 from Senthil Kumar Selvaraj 2013-03-12 17:54:26 UTC --- Created attachment 6931 --> http://sourceware.org/bugzilla/attachment.cgi?id=6931 Patch that loops over all sections -- Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/12494] Relaxation leads to wrong code optimization (computed goto)
http://sourceware.org/bugzilla/show_bug.cgi?id=12494 Senthil Kumar Selvaraj changed: What|Removed |Added AssignedTo|unassigned at sources dot |senthil.thecoder at gmail |redhat.com |dot com --- Comment #10 from Senthil Kumar Selvaraj 2013-03-12 18:03:45 UTC --- Created attachment 6932 --> http://sourceware.org/bugzilla/attachment.cgi?id=6932 Patch that loops over all sections -- Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/12494] Relaxation leads to wrong code optimization (computed goto)
http://sourceware.org/bugzilla/show_bug.cgi?id=12494 Senthil Kumar Selvaraj changed: What|Removed |Added Status|NEW |ASSIGNED CC||senthil.thecoder at gmail ||dot com --- Comment #9 from Senthil Kumar Selvaraj 2013-03-12 18:02:51 UTC --- Actually, the linker relaxation code already considers cases where deletion of the ret in call/ret or jmp/ret might be unsafe - just that it missed this specific case. The case being that GCC puts the array containing the addresses of the labels in a separate section (.rodata). The safe ret deletion logic assumes that only relocations in the local section (containing the ret) could potentially have symbols whose values evaluate to the address containing the ret, but that is not true in this case. Looping over all sections in the BFD, and repeating the check for each section fixes the problem (see attached patch). Changelog 2013-03-12 Senthil Kumar Selvaraj * bfd/elf32-avr.c: Consider all sections to determine if linker relaxation can safely delete a ret after a call/jmp -- Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/15281] New: [avr] DWARF CIE DW_CFA_def_cfa offset incorrect for xmega devices with >128 KB flash
http://sourceware.org/bugzilla/show_bug.cgi?id=15281
Bug #: 15281
Summary: [avr] DWARF CIE DW_CFA_def_cfa offset incorrect for
xmega devices with >128 KB flash
Product: binutils
Version: 2.24 (HEAD)
Status: NEW
Severity: normal
Priority: P2
Component: gas
AssignedTo: unassig...@sourceware.org
ReportedBy: senthil.theco...@gmail.com
Classification: Unclassified
The cfa offset definition in the DWARF CIE for xmegas with flash greater th
an
128 KB is 2, instead of 3. For devices with > 128 KB flash, 3 bytes are pus
hed
on the stack as the return address, and hence the value of SP at the call s
ite
(CFA) will be at an offset of 3.
This works correctly for megas with flash >128 KB, as shown below (note the
value of ofs for DW_CFA_def_cfa
➜ scratch cat test.c
int main()
{
return 0;
}
➜ scratch ~/avr/install/bin/avr-gcc -mmcu=atxmega192a3 -c -g3 t
est.c
➜ scratch ~/avr/install/bin/avr-objdump -Wf test.o
test.o: file format elf32-avr
Contents of the .debug_frame section:
0010 CIE
Version: 1
Augmentation: ""
Code alignment factor: 2
Data alignment factor: -1
Return address column: 36
DW_CFA_def_cfa: r32 ofs 2
DW_CFA_offset: r36 at cfa-1
DW_CFA_nop
DW_CFA_nop
...
➜ scratch ~/avr/install/bin/avr-gcc -mmcu=atmega2560 -c -g3 tes
t.c
➜ scratch ~/avr/install/bin/avr-objdump -Wf test.o
test.o: file format elf32-avr
Contents of the .debug_frame section:
0010 CIE
Version: 1
Augmentation: ""
Code alignment factor: 2
Data alignment factor: -1
Return address column: 36
DW_CFA_def_cfa: r32 ofs 3
DW_CFA_offset: r36 at cfa-2
DW_CFA_nop
DW_CFA_nop
--
Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email
--- You are receiving this mail because: ---
You are on the CC list for the bug.
___
bug-binutils mailing list
bug-binutils@gnu.org
https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/15281] [avr] DWARF CIE DW_CFA_def_cfa offset incorrect for xmega devices with >128 KB flash
http://sourceware.org/bugzilla/show_bug.cgi?id=15281 Senthil Kumar Selvaraj changed: What|Removed |Added Target||avr Host||linux x86_64 --- Comment #1 from Senthil Kumar Selvaraj 2013-03-15 13:34:42 UTC --- The root cause is that tc_cfi_frame_initial_instructions assumes only devices under bfd_mach_avr6 push 3 bytes on the stack as the return address for a function call. bfd_mach_avrxmega6 and bfd_mach_avrxmega7 also have > 128 KB flash and therefore push 3 bytes too. Attaching patch that fixes the problem. -- Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug gas/15281] [avr] DWARF CIE DW_CFA_def_cfa offset incorrect for xmega devices with >128 KB flash
http://sourceware.org/bugzilla/show_bug.cgi?id=15281 --- Comment #2 from Senthil Kumar Selvaraj 2013-03-15 13:35:55 UTC --- Created attachment 6935 --> http://sourceware.org/bugzilla/attachment.cgi?id=6935 Fix for dwarf cie cfa offset for xmega6 and xmega7 -- Configure bugmail: http://sourceware.org/bugzilla/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/20221] New: [avr] Wrong code generated with linker relaxation for source with alignment directives
https://sourceware.org/bugzilla/show_bug.cgi?id=20221 Bug ID: 20221 Summary: [avr] Wrong code generated with linker relaxation for source with alignment directives Product: binutils Version: 2.27 (HEAD) Status: UNCONFIRMED Severity: normal Priority: P2 Component: ld Assignee: unassigned at sourceware dot org Reporter: senthil.thecoder at gmail dot com Target Milestone: --- $ cat avr-prop-5.s .text .global _start, dest _start: CALLdest .align 1 dest: NOP when assembled with -mlink-relax and linked with --relax $ avr-gcc -mrelax -mmcu=atmega1280 -Os avr-prop-5.s -nostartfiles $ avr-objdump -S a.out a.out: file format elf32-avr Disassembly of section .text: <__ctors_end>: 0: ff df rcall .-2 ; 0x0 <__ctors_end> 0002 : ... Note that the rcall offset is wrong - it's calling 0x0 rather than calling dest (0x2). Disabling relaxation results in correct code $ avr-gcc -mno-relax -mmcu=atmega1280 -Os avr-prop-5.s -nostartfiles $ avr-objdump -S a.out a.out: file format elf32-avr Disassembly of section .text: <__ctors_end>: 0: 0e 94 02 00 call0x4 ; 0x4 0004 : ... -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/20221] [avr] Wrong code generated with linker relaxation for source with alignment directives
https://sourceware.org/bugzilla/show_bug.cgi?id=20221 --- Comment #2 from Senthil Kumar Selvaraj --- There's another related testcase that fails even with or without the current fix in master. It fails on binutils-2_26 branch too. .text .global _start, dest _start: jmp dest .align1 dest: nop rjmp dest With $ avr-gcc -mmcu=atmega1280 -Os -mrelax test.s -nostartfiles $ avr-objdump -S a.out <__ctors_end>: 0: 00 c0 rjmp.+0 ; 0x2 0002 : 2: 00 00 nop 4: 00 c0 rjmp.+0 ; 0x6 <_etext> which is obviously wrong. -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/20254] New: [avr] Relocations at end of a section are not processed in the presence of alignment directives
https://sourceware.org/bugzilla/show_bug.cgi?id=20254 Bug ID: 20254 Summary: [avr] Relocations at end of a section are not processed in the presence of alignment directives Product: binutils Version: 2.27 (HEAD) Status: UNCONFIRMED Severity: normal Priority: P2 Component: ld Assignee: unassigned at sourceware dot org Reporter: senthil.thecoder at gmail dot com Target Milestone: --- This fails on binutils-2_26 branch. The fix to PR 20221 does not fix this. $ cat test.s .text .global _start, dest _start: jmp dest .align1 dest: nop rjmp dest With $ avr-gcc -mmcu=atmega1280 -Os -mrelax test.s -nostartfiles $ avr-objdump -S a.out <__ctors_end>: 0: 00 c0 rjmp.+0 ; 0x2 0002 : 2: 00 00 nop 4: 00 c0 rjmp.+0 ; 0x6 <_etext> The rjmp at address 4 jumps to 0x6, instead of 0x2 (which is where dest is). -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/20254] [avr] Relocations at end of a section are not processed in the presence of alignment directives
https://sourceware.org/bugzilla/show_bug.cgi?id=20254 Senthil Kumar Selvaraj changed: What|Removed |Added Target||avr -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
[Bug ld/20221] [avr] Wrong code generated with linker relaxation for source with alignment directives
https://sourceware.org/bugzilla/show_bug.cgi?id=20221 Senthil Kumar Selvaraj changed: What|Removed |Added Target||avr -- You are receiving this mail because: You are on the CC list for the bug. ___ bug-binutils mailing list bug-binutils@gnu.org https://lists.gnu.org/mailman/listinfo/bug-binutils
