Hi everybody,
During the LCU13 cross distro session Riku brought up the topic of aarch64 BE standardization. After consulting with some of our tools team, we propose the following standard:
Linker path: /lib/ld-linux-aarch64be.so.1
Autoconf triplet: aarch64be-*-linux*
Basically, we'd like to tack a be at the end of aarch64. Will this work for everybody?
On Mon, Nov 11, 2013 at 03:57:06PM -0800, Brendan Conoboy wrote:
Hi everybody,
During the LCU13 cross distro session Riku brought up the topic of aarch64 BE standardization. After consulting with some of our tools team, we propose the following standard:
Linker path: /lib/ld-linux-aarch64be.so.1
Autoconf triplet: aarch64be-*-linux*
Basically, we'd like to tack a be at the end of aarch64. Will this work for everybody?
I *think* the linker path works, yes. But there's already a defined triplet that's different, IIRC. Adding CCs to Richard and Matt too for their input as ARM and Linaro toolchain maintainers...
/me tries to resist the urge to bikeshed with "eb" rather than "be" ... :-)
Cheers,
Gcc already uses aarch64_be-*-* for big-endian and has done since the port was upstreamed. Please let's stick to that.
R
On 12 Nov 2013, at 18:37, "Steve McIntyre" steve.mcintyre@linaro.org wrote:
On Mon, Nov 11, 2013 at 03:57:06PM -0800, Brendan Conoboy wrote:
Hi everybody,
During the LCU13 cross distro session Riku brought up the topic of aarch64 BE standardization. After consulting with some of our tools team, we propose the following standard:
Linker path: /lib/ld-linux-aarch64be.so.1
Autoconf triplet: aarch64be-*-linux*
Basically, we'd like to tack a be at the end of aarch64. Will this work for everybody?
I *think* the linker path works, yes. But there's already a defined triplet that's different, IIRC. Adding CCs to Richard and Matt too for their input as ARM and Linaro toolchain maintainers...
/me tries to resist the urge to bikeshed with "eb" rather than "be" ... :-)
Cheers,
Steve McIntyre steve.mcintyre@linaro.org http://www.linaro.org/ Linaro.org | Open source software for ARM SoCs
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
I should add that autoconf was also patched to support that.
On 12 Nov 2013, at 19:07, "Richard Earnshaw" Richard.Earnshaw@arm.com wrote:
Gcc already uses aarch64_be-*-* for big-endian and has done since the port was upstreamed. Please let's stick to that.
R
On 12 Nov 2013, at 18:37, "Steve McIntyre" steve.mcintyre@linaro.org wrote:
On Mon, Nov 11, 2013 at 03:57:06PM -0800, Brendan Conoboy wrote:
Hi everybody,
During the LCU13 cross distro session Riku brought up the topic of aarch64 BE standardization. After consulting with some of our tools team, we propose the following standard:
Linker path: /lib/ld-linux-aarch64be.so.1
Autoconf triplet: aarch64be-*-linux*
Basically, we'd like to tack a be at the end of aarch64. Will this work for everybody?
I *think* the linker path works, yes. But there's already a defined triplet that's different, IIRC. Adding CCs to Richard and Matt too for their input as ARM and Linaro toolchain maintainers...
/me tries to resist the urge to bikeshed with "eb" rather than "be" ... :-)
Cheers,
Steve McIntyre steve.mcintyre@linaro.org http://www.linaro.org/ Linaro.org | Open source software for ARM SoCs
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
On 11/12/2013 02:06 PM, Richard Earnshaw wrote:
Gcc already uses aarch64_be-*-* for big-endian and has done since the port was upstreamed. Please let's stick to that.
I assume then that the unique and logical choice for the dyanmic linker is:
/lib/ld-linux-aarch64_be.so.1
Shall we all agree to this?
As Steve pointed out it's entirely bike-shed, we just want everyone on the same page.
I warn everyone that this is the first dynamic linker with an underscore in the name.
I don't know exactly what impact that will have on regexps used by other tools, but we can just clean things up as we see them (note x86_64 is: /lib64/ld-linux-x86-64.so.2)
Cheers, Carlos.
Makes sense to me.
R.
On 12 Nov 2013, at 20:13, "Carlos O'Donell" carlos@redhat.com wrote:
On 11/12/2013 02:06 PM, Richard Earnshaw wrote:
Gcc already uses aarch64_be-*-* for big-endian and has done since the port was upstreamed. Please let's stick to that.
I assume then that the unique and logical choice for the dyanmic linker is:
/lib/ld-linux-aarch64_be.so.1
Shall we all agree to this?
As Steve pointed out it's entirely bike-shed, we just want everyone on the same page.
I warn everyone that this is the first dynamic linker with an underscore in the name.
I don't know exactly what impact that will have on regexps used by other tools, but we can just clean things up as we see them (note x86_64 is: /lib64/ld-linux-x86-64.so.2)
Cheers, Carlos.
-- IMPORTANT NOTICE: The contents of this email and any attachments are confidential and may also be privileged. If you are not the intended recipient, please notify the sender immediately and do not disclose the contents to any other person, use it for any purpose, or store or copy the information in any medium. Thank you.
ARM Limited, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2557590 ARM Holdings plc, Registered office 110 Fulbourn Road, Cambridge CB1 9NJ, Registered in England & Wales, Company No: 2548782
On Tue, Nov 12, 2013 at 02:16:13PM -0500, Carlos O'Donell wrote:
On 11/12/2013 02:06 PM, Richard Earnshaw wrote:
Gcc already uses aarch64_be-*-* for big-endian and has done since the port was upstreamed. Please let's stick to that.
I assume then that the unique and logical choice for the dyanmic linker is:
/lib/ld-linux-aarch64_be.so.1
Shall we all agree to this?
Makes sense to me, yup.
As Steve pointed out it's entirely bike-shed, we just want everyone on the same page.
I warn everyone that this is the first dynamic linker with an underscore in the name.
I don't know exactly what impact that will have on regexps used by other tools, but we can just clean things up as we see them (note x86_64 is: /lib64/ld-linux-x86-64.so.2)
Suck it and see, yes.
Cheers,
On 11/12/2013 04:22 PM, Steve McIntyre wrote:
On Tue, Nov 12, 2013 at 02:16:13PM -0500, Carlos O'Donell wrote:
On 11/12/2013 02:06 PM, Richard Earnshaw wrote:
Gcc already uses aarch64_be-*-* for big-endian and has done since the port was upstreamed. Please let's stick to that.
I assume then that the unique and logical choice for the dyanmic linker is:
/lib/ld-linux-aarch64_be.so.1
Shall we all agree to this?
Makes sense to me, yup.
As Steve pointed out it's entirely bike-shed, we just want everyone on the same page.
I warn everyone that this is the first dynamic linker with an underscore in the name.
I don't know exactly what impact that will have on regexps used by other tools, but we can just clean things up as we see them (note x86_64 is: /lib64/ld-linux-x86-64.so.2)
Suck it and see, yes.
That looks like consensus to me.
We'll get upstream glibc to use /lib/ld-linux-aarch64_be.so.1 for aarch64_be-*-* on Linux targets.
I assume that there will be no other ABI variant dynamic linkers for AArch64 and if there are we'll need to do this process again and pick another unique name if we want to support a mixed environment.
Steve,
Do you know if anyone at Linaro has glibc patches for AArch64 BE?
Cheers, Carlos.
-
Brendan Conoboy -
Carlos O'Donell -
Richard Earnshaw -
Steve McIntyre