[U-Boot] [PATCH] arm: lds: Remove libgcc eabi exception handling tables
Remove ARM eabi exception handling tables (for frame unwinding). AFAICT, u-boot stubs away the frame unwiding routines, so the tables will more or less just consume space. It should be OK to remove them.
Signed-off-by: Edgar E. Iglesias edgar.iglesias@xilinx.com Signed-off-by: Michal Simek michal.simek@xilinx.com --- This patch was sent to ML as RFC (May 9)
Here is the origin response in connection to this patch.
Ok, so Michal and I just did some fiddling with zynq builds and *exidx* sections.
By default the *exidx* sections are between rodata and data, so removing them causes many apparent changes at the binary level. However, builds of zynq based on ARM master with the patch above vs master with a patch mapping *exidx* sections after BSS gives identical binaries. Thus the RFC has no functional effect.
Also, ARM EHABI states that [exception] Tables are not required for ABI compliance at the C/Assembler level but are required for C++.
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038a/IHI0038A_ehabi.pdf
So as long as we don't put any C++ code in U-Boot (a prospect that I don't see happening any time soon), this RFC is safe and either is a no-op or removes useless bytes from the binary.
--- arch/arm/cpu/u-boot.lds | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/arch/arm/cpu/u-boot.lds b/arch/arm/cpu/u-boot.lds index 3037885..8894c8a 100644 --- a/arch/arm/cpu/u-boot.lds +++ b/arch/arm/cpu/u-boot.lds @@ -113,4 +113,6 @@ SECTIONS /DISCARD/ : { *(.plt*) } /DISCARD/ : { *(.interp*) } /DISCARD/ : { *(.gnu*) } + /DISCARD/ : { *(.ARM.exidx*) } + /DISCARD/ : { *(.gnu.linkonce.armexidx.*) } } -- 1.8.2.3
Hi Michal,
On Thu, 25 Jul 2013 16:16:46 +0200, Michal Simek michal.simek@xilinx.com wrote:
Remove ARM eabi exception handling tables (for frame unwinding). AFAICT, u-boot stubs away the frame unwiding routines, so the tables will more or less just consume space. It should be OK to remove them.
Signed-off-by: Edgar E. Iglesias edgar.iglesias@xilinx.com Signed-off-by: Michal Simek michal.simek@xilinx.com
This patch was sent to ML as RFC (May 9)
Here is the origin response in connection to this patch.
Ok, so Michal and I just did some fiddling with zynq builds and *exidx* sections.
By default the *exidx* sections are between rodata and data, so removing them causes many apparent changes at the binary level. However, builds of zynq based on ARM master with the patch above vs master with a patch mapping *exidx* sections after BSS gives identical binaries. Thus the RFC has no functional effect.
Also, ARM EHABI states that [exception] Tables are not required for ABI compliance at the C/Assembler level but are required for C++.
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038a/IHI0038A_ehabi.pdf
So as long as we don't put any C++ code in U-Boot (a prospect that I don't see happening any time soon), this RFC is safe and either is a no-op or removes useless bytes from the binary.
arch/arm/cpu/u-boot.lds | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/arch/arm/cpu/u-boot.lds b/arch/arm/cpu/u-boot.lds index 3037885..8894c8a 100644 --- a/arch/arm/cpu/u-boot.lds +++ b/arch/arm/cpu/u-boot.lds @@ -113,4 +113,6 @@ SECTIONS /DISCARD/ : { *(.plt*) } /DISCARD/ : { *(.interp*) } /DISCARD/ : { *(.gnu*) }
- /DISCARD/ : { *(.ARM.exidx*) }
- /DISCARD/ : { *(.gnu.linkonce.armexidx.*) }
}
1.8.2.3
The good news is, it applies and does not cause any build issues.
The less good news is, arch/arm/cpu/u-boot.lds is not the only linker script file to fix if we want to fix it this way.
The fact is, I am not sure removing sections from the ELF filein order to avoid them being in the binary is actually the right way to go. I have recently had to 'un-discard' some sections in order for 'offline' debugging tools to properly understand the ELF file.
So I am wondering if we should not stop using /DISCARD/ and start using $(CROSS_COMPILE)objcopy options (-j or --strip-* and --remove-section).
This way, the ELF file would be as complete as any debug tool might expect it (e.g. objdump, especially when debugging relocation issues) yet the binary file would remain unchanged.
But past rc2 is not a time to start such a change.
Still, I'd like your fix to be consistent across all of ARM. Can you change all the linker scripts used for build ARM targets? These are, according to an ugly mix of cat LOG/*.MAKELOG, sed, grep and sort -u:
arch/arm/cpu/arm1136/u-boot-spl.lds arch/arm/cpu/arm926ejs/mxs/u-boot-spl.lds arch/arm/cpu/arm926ejs/spear/u-boot-spl.lds arch/arm/cpu/armv7/am33xx/u-boot-spl.lds arch/arm/cpu/armv7/omap-common/u-boot-spl.lds arch/arm/cpu/armv7/socfpga/u-boot-spl.lds arch/arm/cpu/ixp/u-boot.lds arch/arm/cpu/u-boot.lds arch/arm/cpu/u-boot-spl.lds board/actux1/u-boot.lds board/actux2/u-boot.lds board/actux3/u-boot.lds board/ait/cam_enc_4xx/u-boot-spl.lds board/davinci/da8xxevm/u-boot-spl-da850evm.lds board/davinci/da8xxevm/u-boot-spl-hawk.lds board/dvlhost/u-boot.lds board/freescale/mx31ads/u-boot.lds board/samsung/common/exynos-uboot-spl.lds board/ti/am335x/u-boot.lds board/vpac270/u-boot-spl.lds
Amicalement,
Hi Albert,
On 09/05/2013 03:03 PM, Albert ARIBAUD wrote:
Hi Michal,
On Thu, 25 Jul 2013 16:16:46 +0200, Michal Simek michal.simek@xilinx.com wrote:
Remove ARM eabi exception handling tables (for frame unwinding). AFAICT, u-boot stubs away the frame unwiding routines, so the tables will more or less just consume space. It should be OK to remove them.
Signed-off-by: Edgar E. Iglesias edgar.iglesias@xilinx.com Signed-off-by: Michal Simek michal.simek@xilinx.com
This patch was sent to ML as RFC (May 9)
Here is the origin response in connection to this patch.
Ok, so Michal and I just did some fiddling with zynq builds and *exidx* sections.
By default the *exidx* sections are between rodata and data, so removing them causes many apparent changes at the binary level. However, builds of zynq based on ARM master with the patch above vs master with a patch mapping *exidx* sections after BSS gives identical binaries. Thus the RFC has no functional effect.
Also, ARM EHABI states that [exception] Tables are not required for ABI compliance at the C/Assembler level but are required for C++.
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038a/IHI0038A_ehabi.pdf
So as long as we don't put any C++ code in U-Boot (a prospect that I don't see happening any time soon), this RFC is safe and either is a no-op or removes useless bytes from the binary.
arch/arm/cpu/u-boot.lds | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/arch/arm/cpu/u-boot.lds b/arch/arm/cpu/u-boot.lds index 3037885..8894c8a 100644 --- a/arch/arm/cpu/u-boot.lds +++ b/arch/arm/cpu/u-boot.lds @@ -113,4 +113,6 @@ SECTIONS /DISCARD/ : { *(.plt*) } /DISCARD/ : { *(.interp*) } /DISCARD/ : { *(.gnu*) }
- /DISCARD/ : { *(.ARM.exidx*) }
- /DISCARD/ : { *(.gnu.linkonce.armexidx.*) }
}
1.8.2.3
The good news is, it applies and does not cause any build issues.
The less good news is, arch/arm/cpu/u-boot.lds is not the only linker script file to fix if we want to fix it this way.
The fact is, I am not sure removing sections from the ELF filein order to avoid them being in the binary is actually the right way to go. I have recently had to 'un-discard' some sections in order for 'offline' debugging tools to properly understand the ELF file.
So I am wondering if we should not stop using /DISCARD/ and start using $(CROSS_COMPILE)objcopy options (-j or --strip-* and --remove-section).
This way, the ELF file would be as complete as any debug tool might expect it (e.g. objdump, especially when debugging relocation issues) yet the binary file would remain unchanged.
But past rc2 is not a time to start such a change.
Still, I'd like your fix to be consistent across all of ARM. Can you change all the linker scripts used for build ARM targets? These are, according to an ugly mix of cat LOG/*.MAKELOG, sed, grep and sort -u:
arch/arm/cpu/arm1136/u-boot-spl.lds arch/arm/cpu/arm926ejs/mxs/u-boot-spl.lds arch/arm/cpu/arm926ejs/spear/u-boot-spl.lds arch/arm/cpu/armv7/am33xx/u-boot-spl.lds arch/arm/cpu/armv7/omap-common/u-boot-spl.lds arch/arm/cpu/armv7/socfpga/u-boot-spl.lds arch/arm/cpu/ixp/u-boot.lds arch/arm/cpu/u-boot.lds arch/arm/cpu/u-boot-spl.lds board/actux1/u-boot.lds board/actux2/u-boot.lds board/actux3/u-boot.lds board/ait/cam_enc_4xx/u-boot-spl.lds board/davinci/da8xxevm/u-boot-spl-da850evm.lds board/davinci/da8xxevm/u-boot-spl-hawk.lds board/dvlhost/u-boot.lds board/freescale/mx31ads/u-boot.lds board/samsung/common/exynos-uboot-spl.lds board/ti/am335x/u-boot.lds board/vpac270/u-boot-spl.lds
Honestly for me is much easier just fix our timer code not to use this section instead of trying to fix 3rd party platforms. Is there any reason why too many linkers scripts are in u-boot for ARM?
Thanks, Michal
Hi Michal,
On Thu, 05 Sep 2013 17:30:54 +0200, Michal Simek monstr@monstr.eu wrote:
Hi Albert,
On 09/05/2013 03:03 PM, Albert ARIBAUD wrote:
Hi Michal,
On Thu, 25 Jul 2013 16:16:46 +0200, Michal Simek michal.simek@xilinx.com wrote:
Remove ARM eabi exception handling tables (for frame unwinding). AFAICT, u-boot stubs away the frame unwiding routines, so the tables will more or less just consume space. It should be OK to remove them.
Signed-off-by: Edgar E. Iglesias edgar.iglesias@xilinx.com Signed-off-by: Michal Simek michal.simek@xilinx.com
This patch was sent to ML as RFC (May 9)
Here is the origin response in connection to this patch.
Ok, so Michal and I just did some fiddling with zynq builds and *exidx* sections.
By default the *exidx* sections are between rodata and data, so removing them causes many apparent changes at the binary level. However, builds of zynq based on ARM master with the patch above vs master with a patch mapping *exidx* sections after BSS gives identical binaries. Thus the RFC has no functional effect.
Also, ARM EHABI states that [exception] Tables are not required for ABI compliance at the C/Assembler level but are required for C++.
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038a/IHI0038A_ehabi.pdf
So as long as we don't put any C++ code in U-Boot (a prospect that I don't see happening any time soon), this RFC is safe and either is a no-op or removes useless bytes from the binary.
arch/arm/cpu/u-boot.lds | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/arch/arm/cpu/u-boot.lds b/arch/arm/cpu/u-boot.lds index 3037885..8894c8a 100644 --- a/arch/arm/cpu/u-boot.lds +++ b/arch/arm/cpu/u-boot.lds @@ -113,4 +113,6 @@ SECTIONS /DISCARD/ : { *(.plt*) } /DISCARD/ : { *(.interp*) } /DISCARD/ : { *(.gnu*) }
- /DISCARD/ : { *(.ARM.exidx*) }
- /DISCARD/ : { *(.gnu.linkonce.armexidx.*) }
}
1.8.2.3
The good news is, it applies and does not cause any build issues.
The less good news is, arch/arm/cpu/u-boot.lds is not the only linker script file to fix if we want to fix it this way.
The fact is, I am not sure removing sections from the ELF filein order to avoid them being in the binary is actually the right way to go. I have recently had to 'un-discard' some sections in order for 'offline' debugging tools to properly understand the ELF file.
So I am wondering if we should not stop using /DISCARD/ and start using $(CROSS_COMPILE)objcopy options (-j or --strip-* and --remove-section).
This way, the ELF file would be as complete as any debug tool might expect it (e.g. objdump, especially when debugging relocation issues) yet the binary file would remain unchanged.
But past rc2 is not a time to start such a change.
Still, I'd like your fix to be consistent across all of ARM. Can you change all the linker scripts used for build ARM targets? These are, according to an ugly mix of cat LOG/*.MAKELOG, sed, grep and sort -u:
arch/arm/cpu/arm1136/u-boot-spl.lds arch/arm/cpu/arm926ejs/mxs/u-boot-spl.lds arch/arm/cpu/arm926ejs/spear/u-boot-spl.lds arch/arm/cpu/armv7/am33xx/u-boot-spl.lds arch/arm/cpu/armv7/omap-common/u-boot-spl.lds arch/arm/cpu/armv7/socfpga/u-boot-spl.lds arch/arm/cpu/ixp/u-boot.lds arch/arm/cpu/u-boot.lds arch/arm/cpu/u-boot-spl.lds board/actux1/u-boot.lds board/actux2/u-boot.lds board/actux3/u-boot.lds board/ait/cam_enc_4xx/u-boot-spl.lds board/davinci/da8xxevm/u-boot-spl-da850evm.lds board/davinci/da8xxevm/u-boot-spl-hawk.lds board/dvlhost/u-boot.lds board/freescale/mx31ads/u-boot.lds board/samsung/common/exynos-uboot-spl.lds board/ti/am335x/u-boot.lds board/vpac270/u-boot-spl.lds
Honestly for me is much easier just fix our timer code not to use this section instead of trying to fix 3rd party platforms.
I was only asking. I did not intend to force any work upon you. As I said, we're past -rc2.
So, I'll just pick it as-is. Will your timer code go into mainline soon? This would ensure that a build failure would be detected immediately if the change was reverted by any later commit such as the "make a full ELF and a slashed .bin" change I have mentioned.
Is there any reason why too many linkers scripts are in u-boot for ARM?
Yes: copy-paste. :)
More seriously, when start.o files started popping in different places for different ARM targets, this was handled by duplicating .lds files and changing the start.o location in each .lds. Then it got worse when some targets starting needing different memory region mappings.
I am working on reducing both the start.o and lds mess. I was initially aiming for 2013.10, will do it for 2014.01.
Thanks, Michal
Amicalement,
Hi,
On 09/05/2013 06:05 PM, Albert ARIBAUD wrote:
Hi Michal,
On Thu, 05 Sep 2013 17:30:54 +0200, Michal Simek monstr@monstr.eu wrote:
Hi Albert,
On 09/05/2013 03:03 PM, Albert ARIBAUD wrote:
Hi Michal,
On Thu, 25 Jul 2013 16:16:46 +0200, Michal Simek michal.simek@xilinx.com wrote:
Remove ARM eabi exception handling tables (for frame unwinding). AFAICT, u-boot stubs away the frame unwiding routines, so the tables will more or less just consume space. It should be OK to remove them.
Signed-off-by: Edgar E. Iglesias edgar.iglesias@xilinx.com Signed-off-by: Michal Simek michal.simek@xilinx.com
This patch was sent to ML as RFC (May 9)
Here is the origin response in connection to this patch.
Ok, so Michal and I just did some fiddling with zynq builds and *exidx* sections.
By default the *exidx* sections are between rodata and data, so removing them causes many apparent changes at the binary level. However, builds of zynq based on ARM master with the patch above vs master with a patch mapping *exidx* sections after BSS gives identical binaries. Thus the RFC has no functional effect.
Also, ARM EHABI states that [exception] Tables are not required for ABI compliance at the C/Assembler level but are required for C++.
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038a/IHI0038A_ehabi.pdf
So as long as we don't put any C++ code in U-Boot (a prospect that I don't see happening any time soon), this RFC is safe and either is a no-op or removes useless bytes from the binary.
arch/arm/cpu/u-boot.lds | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/arch/arm/cpu/u-boot.lds b/arch/arm/cpu/u-boot.lds index 3037885..8894c8a 100644 --- a/arch/arm/cpu/u-boot.lds +++ b/arch/arm/cpu/u-boot.lds @@ -113,4 +113,6 @@ SECTIONS /DISCARD/ : { *(.plt*) } /DISCARD/ : { *(.interp*) } /DISCARD/ : { *(.gnu*) }
- /DISCARD/ : { *(.ARM.exidx*) }
- /DISCARD/ : { *(.gnu.linkonce.armexidx.*) }
}
1.8.2.3
The good news is, it applies and does not cause any build issues.
The less good news is, arch/arm/cpu/u-boot.lds is not the only linker script file to fix if we want to fix it this way.
The fact is, I am not sure removing sections from the ELF filein order to avoid them being in the binary is actually the right way to go. I have recently had to 'un-discard' some sections in order for 'offline' debugging tools to properly understand the ELF file.
So I am wondering if we should not stop using /DISCARD/ and start using $(CROSS_COMPILE)objcopy options (-j or --strip-* and --remove-section).
This way, the ELF file would be as complete as any debug tool might expect it (e.g. objdump, especially when debugging relocation issues) yet the binary file would remain unchanged.
But past rc2 is not a time to start such a change.
Still, I'd like your fix to be consistent across all of ARM. Can you change all the linker scripts used for build ARM targets? These are, according to an ugly mix of cat LOG/*.MAKELOG, sed, grep and sort -u:
arch/arm/cpu/arm1136/u-boot-spl.lds arch/arm/cpu/arm926ejs/mxs/u-boot-spl.lds arch/arm/cpu/arm926ejs/spear/u-boot-spl.lds arch/arm/cpu/armv7/am33xx/u-boot-spl.lds arch/arm/cpu/armv7/omap-common/u-boot-spl.lds arch/arm/cpu/armv7/socfpga/u-boot-spl.lds arch/arm/cpu/ixp/u-boot.lds arch/arm/cpu/u-boot.lds arch/arm/cpu/u-boot-spl.lds board/actux1/u-boot.lds board/actux2/u-boot.lds board/actux3/u-boot.lds board/ait/cam_enc_4xx/u-boot-spl.lds board/davinci/da8xxevm/u-boot-spl-da850evm.lds board/davinci/da8xxevm/u-boot-spl-hawk.lds board/dvlhost/u-boot.lds board/freescale/mx31ads/u-boot.lds board/samsung/common/exynos-uboot-spl.lds board/ti/am335x/u-boot.lds board/vpac270/u-boot-spl.lds
Honestly for me is much easier just fix our timer code not to use this section instead of trying to fix 3rd party platforms.
I was only asking. I did not intend to force any work upon you. As I said, we're past -rc2.
So, I'll just pick it as-is. Will your timer code go into mainline soon? This would ensure that a build failure would be detected immediately if the change was reverted by any later commit such as the "make a full ELF and a slashed .bin" change I have mentioned.
The timer code is there but only in one configuration there is this issue. It is not causing build failure. It is just when you create xilinx specific boot image then you face with this problem.
Is there any reason why too many linkers scripts are in u-boot for ARM?
Yes: copy-paste. :)
:-)
More seriously, when start.o files started popping in different places for different ARM targets, this was handled by duplicating .lds files and changing the start.o location in each .lds. Then it got worse when some targets starting needing different memory region mappings.
especially because of SPL.
I am working on reducing both the start.o and lds mess. I was initially aiming for 2013.10, will do it for 2014.01.
Great.
Thanks, Michal
Hi Michal,
On Thu, 25 Jul 2013 16:16:46 +0200, Michal Simek michal.simek@xilinx.com wrote:
Remove ARM eabi exception handling tables (for frame unwinding). AFAICT, u-boot stubs away the frame unwiding routines, so the tables will more or less just consume space. It should be OK to remove them.
Signed-off-by: Edgar E. Iglesias edgar.iglesias@xilinx.com Signed-off-by: Michal Simek michal.simek@xilinx.com
This patch was sent to ML as RFC (May 9)
Here is the origin response in connection to this patch.
Ok, so Michal and I just did some fiddling with zynq builds and *exidx* sections.
By default the *exidx* sections are between rodata and data, so removing them causes many apparent changes at the binary level. However, builds of zynq based on ARM master with the patch above vs master with a patch mapping *exidx* sections after BSS gives identical binaries. Thus the RFC has no functional effect.
Also, ARM EHABI states that [exception] Tables are not required for ABI compliance at the C/Assembler level but are required for C++.
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038a/IHI0038A_ehabi.pdf
So as long as we don't put any C++ code in U-Boot (a prospect that I don't see happening any time soon), this RFC is safe and either is a no-op or removes useless bytes from the binary.
arch/arm/cpu/u-boot.lds | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/arch/arm/cpu/u-boot.lds b/arch/arm/cpu/u-boot.lds index 3037885..8894c8a 100644 --- a/arch/arm/cpu/u-boot.lds +++ b/arch/arm/cpu/u-boot.lds @@ -113,4 +113,6 @@ SECTIONS /DISCARD/ : { *(.plt*) } /DISCARD/ : { *(.interp*) } /DISCARD/ : { *(.gnu*) }
- /DISCARD/ : { *(.ARM.exidx*) }
- /DISCARD/ : { *(.gnu.linkonce.armexidx.*) }
}
1.8.2.3
Applied to u-boot-arm/master, thanks!
Amicalement,
On 09/05/2013 09:16 PM, Albert ARIBAUD wrote:
Hi Michal,
On Thu, 25 Jul 2013 16:16:46 +0200, Michal Simek michal.simek@xilinx.com wrote:
Remove ARM eabi exception handling tables (for frame unwinding). AFAICT, u-boot stubs away the frame unwiding routines, so the tables will more or less just consume space. It should be OK to remove them.
Signed-off-by: Edgar E. Iglesias edgar.iglesias@xilinx.com Signed-off-by: Michal Simek michal.simek@xilinx.com
This patch was sent to ML as RFC (May 9)
Here is the origin response in connection to this patch.
Ok, so Michal and I just did some fiddling with zynq builds and *exidx* sections.
By default the *exidx* sections are between rodata and data, so removing them causes many apparent changes at the binary level. However, builds of zynq based on ARM master with the patch above vs master with a patch mapping *exidx* sections after BSS gives identical binaries. Thus the RFC has no functional effect.
Also, ARM EHABI states that [exception] Tables are not required for ABI compliance at the C/Assembler level but are required for C++.
http://infocenter.arm.com/help/topic/com.arm.doc.ihi0038a/IHI0038A_ehabi.pdf
So as long as we don't put any C++ code in U-Boot (a prospect that I don't see happening any time soon), this RFC is safe and either is a no-op or removes useless bytes from the binary.
arch/arm/cpu/u-boot.lds | 2 ++ 1 file changed, 2 insertions(+)
diff --git a/arch/arm/cpu/u-boot.lds b/arch/arm/cpu/u-boot.lds index 3037885..8894c8a 100644 --- a/arch/arm/cpu/u-boot.lds +++ b/arch/arm/cpu/u-boot.lds @@ -113,4 +113,6 @@ SECTIONS /DISCARD/ : { *(.plt*) } /DISCARD/ : { *(.interp*) } /DISCARD/ : { *(.gnu*) }
- /DISCARD/ : { *(.ARM.exidx*) }
- /DISCARD/ : { *(.gnu.linkonce.armexidx.*) }
}
1.8.2.3
Applied to u-boot-arm/master, thanks!
Ok. Thanks, Michal
participants (3)
-
Albert ARIBAUD -
Michal Simek -
Michal Simek