Skip to contentSkip to conversation

clang: do not enable live-patching support

PATCH In review Roger Pau Monne 7 participants 12 messages Nov 27, 2019 – Dec 6, 2019

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 clang: do not enable live-patching support

Live-patching requires unique symbols, and sadly the clang build generates a lot of duplicate symbols:

Duplicate symbol 'asid.c#get_cpu_info' (ffff82d0803032c0 != ffff82d0802e0f50) Duplicate symbol 'asid.c#get_cpu_info_from_stack' (ffff82d0802e1080 != ffff82d0803032f0) Duplicate symbol 'ats.c#__list_add' (ffff82d080260a00 != ffff82d080267c70) Duplicate symbol 'boot.c#constant_test_bit' (ffff82d08040ea60 != ffff82d0804372f0) Duplicate symbol 'common.c#clear_bit' (ffff82d080332440 != ffff82d0802d33b0) Duplicate symbol 'common.c#constant_test_bit' (ffff82d080332340 != ffff82d0802d2220) Duplicate symbol 'common.c#cpumask_check' (ffff82d0802d3370 != ffff82d080337b60) Duplicate symbol 'common.c#get_cpu_info' (ffff82d0802d22b0 != ffff82d080331590) Duplicate symbol 'common.c#get_cpu_info_from_stack' (ffff82d0802d31c0 != ffff82d0803374b0) Duplicate symbol 'common.c#pfn_to_pdx' (ffff82d0802d3270 != ffff82d080331e00) Duplicate symbol 'common.c#test_and_set_bit' (ffff82d0802d3360 != ffff82d080332250) Duplicate symbol 'common.c#variable_clear_bit' (ffff82d0802d2270 != ffff82d080337b50) Duplicate symbol 'compat.c#get_cpu_info' (ffff82d08026eab0 != ffff82d080200460) Duplicate symbol 'compat.c#get_cpu_info_from_stack' (ffff82d08026ebd0 != ffff82d080200f70) Duplicate symbol 'cpu_idle.c#get_cpu_info' (ffff82d0802ccb00 != ffff82d08035fcc0) [...]

For the time being disable live-patching when building with clang, since duplicate symbols will trigger a build failure because ENFORCE_UNIQUE_SYMBOLS is now also enabled by default in conjunction with live-patching.

---
Cc: Jürgen Groß <[email protected]>

---
 Config.mk          | 2 ++
 xen/common/Kconfig | 6 +++++-
 2 files changed, 7 insertions(+), 1 deletion(-)
Config.mk +2 -0
index d8f90d75b3..009abda225 100644
@@ -157,6 +157,8 @@ ifndef XEN_HAS_CHECKPOLICY
     export XEN_HAS_CHECKPOLICY
 endif
 
+export XEN_BUILD_WITH_CLANG = $(clang)
+
 # as-insn: Check whether assembler supports an instruction.
 # Usage: cflags-y += $(call as-insn,CC FLAGS,"insn",option-yes,option-no)
 as-insn = $(if $(shell echo 'void _(void) { asm volatile ( $(2) ); }' \
xen/common/Kconfig +5 -1
index f754741972..097996fc6c 100644
@@ -80,6 +80,10 @@ config HAS_CHECKPOLICY
 	string
 	option env="XEN_HAS_CHECKPOLICY"
 
+config BUILD_WITH_CLANG
+	string
+	option env="XEN_BUILD_WITH_CLANG"
+
 menu "Speculative hardening"
 
 config SPECULATIVE_HARDEN_ARRAY
@@ -350,7 +354,7 @@ config CRYPTO
 config LIVEPATCH
 	bool "Live patching support"
 	default X86
-	depends on HAS_BUILD_ID = "y"
+	depends on HAS_BUILD_ID = "y" && BUILD_WITH_CLANG != "y"
 	---help---
 	  Allows a running Xen hypervisor to be dynamically patched using
 	  binary patches without rebooting. This is primarily used to binarily
Signature · 7 lines
2.24.0


_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel
#1 lore raw
Quoted 20 lines · Roger Pau Monne

Live-patching requires unique symbols, and sadly the clang build generates a lot of duplicate symbols:

Duplicate symbol 'asid.c#get_cpu_info' (ffff82d0803032c0 != ffff82d0802e0f50) Duplicate symbol 'asid.c#get_cpu_info_from_stack' (ffff82d0802e1080 != ffff82d0803032f0) Duplicate symbol 'ats.c#__list_add' (ffff82d080260a00 != ffff82d080267c70) Duplicate symbol 'boot.c#constant_test_bit' (ffff82d08040ea60 != ffff82d0804372f0) Duplicate symbol 'common.c#clear_bit' (ffff82d080332440 != ffff82d0802d33b0) Duplicate symbol 'common.c#constant_test_bit' (ffff82d080332340 != ffff82d0802d2220) Duplicate symbol 'common.c#cpumask_check' (ffff82d0802d3370 != ffff82d080337b60) Duplicate symbol 'common.c#get_cpu_info' (ffff82d0802d22b0 != ffff82d080331590) Duplicate symbol 'common.c#get_cpu_info_from_stack' (ffff82d0802d31c0 != ffff82d0803374b0) Duplicate symbol 'common.c#pfn_to_pdx' (ffff82d0802d3270 != ffff82d080331e00) Duplicate symbol 'common.c#test_and_set_bit' (ffff82d0802d3360 != ffff82d080332250) Duplicate symbol 'common.c#variable_clear_bit' (ffff82d0802d2270 != ffff82d080337b50) Duplicate symbol 'compat.c#get_cpu_info' (ffff82d08026eab0 != ffff82d080200460) Duplicate symbol 'compat.c#get_cpu_info_from_stack' (ffff82d08026ebd0 != ffff82d080200f70) Duplicate symbol 'cpu_idle.c#get_cpu_info' (ffff82d0802ccb00 != ffff82d08035fcc0) [...]

Quoted 6 lines

For the time being disable live-patching when building with clang, since duplicate symbols will trigger a build failure because ENFORCE_UNIQUE_SYMBOLS is now also enabled by default in conjunction with live-patching.

To be honest, as indicated before I'm inclined to nak this patch on the basis that a proper solution has been posted almost 3 weeks ago (and this was already v2). Nevertheless a remark here:

xen/common/Kconfig config HAS_CHECKPOLICY · line 85

Quoted 9 lines
xen/common/Kconfig +3 -0
@@ -80,6 +80,10 @@ config HAS_CHECKPOLICY
 	string
 	option env="XEN_HAS_CHECKPOLICY"
 
+config BUILD_WITH_CLANG
+	string
+	option env="XEN_BUILD_WITH_CLANG"

Instead of introducing a new option here, ...

xen/common/Kconfig config CRYPTO · line 357

Quoted 6 lines
xen/common/Kconfig +1 -1
@@ -350,7 +354,7 @@ config CRYPTO
 config LIVEPATCH
 	bool "Live patching support"
 	default X86
-	depends on HAS_BUILD_ID = "y"
+	depends on HAS_BUILD_ID = "y" && BUILD_WITH_CLANG != "y"

... seeing this, why don't you simply suppress HAS_BUILD_ID acquiring a value of y in ./Config.mk (accompanied by a suitable comment)?

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#2 lore raw
Quoted 26 lines · Jan Beulich
Quoted 25 lines · Roger Pau Monne

Live-patching requires unique symbols, and sadly the clang build generates a lot of duplicate symbols:

Duplicate symbol 'asid.c#get_cpu_info' (ffff82d0803032c0 != ffff82d0802e0f50) Duplicate symbol 'asid.c#get_cpu_info_from_stack' (ffff82d0802e1080 != ffff82d0803032f0) Duplicate symbol 'ats.c#__list_add' (ffff82d080260a00 != ffff82d080267c70) Duplicate symbol 'boot.c#constant_test_bit' (ffff82d08040ea60 != ffff82d0804372f0) Duplicate symbol 'common.c#clear_bit' (ffff82d080332440 != ffff82d0802d33b0) Duplicate symbol 'common.c#constant_test_bit' (ffff82d080332340 != ffff82d0802d2220) Duplicate symbol 'common.c#cpumask_check' (ffff82d0802d3370 != ffff82d080337b60) Duplicate symbol 'common.c#get_cpu_info' (ffff82d0802d22b0 != ffff82d080331590) Duplicate symbol 'common.c#get_cpu_info_from_stack' (ffff82d0802d31c0 != ffff82d0803374b0) Duplicate symbol 'common.c#pfn_to_pdx' (ffff82d0802d3270 != ffff82d080331e00) Duplicate symbol 'common.c#test_and_set_bit' (ffff82d0802d3360 != ffff82d080332250) Duplicate symbol 'common.c#variable_clear_bit' (ffff82d0802d2270 != ffff82d080337b50) Duplicate symbol 'compat.c#get_cpu_info' (ffff82d08026eab0 != ffff82d080200460) Duplicate symbol 'compat.c#get_cpu_info_from_stack' (ffff82d08026ebd0 != ffff82d080200f70) Duplicate symbol 'cpu_idle.c#get_cpu_info' (ffff82d0802ccb00 != ffff82d08035fcc0) [...]

For the time being disable live-patching when building with clang, since duplicate symbols will trigger a build failure because ENFORCE_UNIQUE_SYMBOLS is now also enabled by default in conjunction with live-patching.

Quoted 5 lines
Quoted 1 line

To be honest, as indicated before I'm inclined to nak this patch on the basis that a proper solution has been posted almost 3 weeks ago (and this was already v2).

What's that patch waiting on?

-George

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#3 lore raw
Quoted 27 lines · George Dunlap
Quoted 26 lines · Jan Beulich
Quoted 25 lines · Roger Pau Monne

Live-patching requires unique symbols, and sadly the clang build generates a lot of duplicate symbols:

Duplicate symbol 'asid.c#get_cpu_info' (ffff82d0803032c0 != ffff82d0802e0f50) Duplicate symbol 'asid.c#get_cpu_info_from_stack' (ffff82d0802e1080 != ffff82d0803032f0) Duplicate symbol 'ats.c#__list_add' (ffff82d080260a00 != ffff82d080267c70) Duplicate symbol 'boot.c#constant_test_bit' (ffff82d08040ea60 != ffff82d0804372f0) Duplicate symbol 'common.c#clear_bit' (ffff82d080332440 != ffff82d0802d33b0) Duplicate symbol 'common.c#constant_test_bit' (ffff82d080332340 != ffff82d0802d2220) Duplicate symbol 'common.c#cpumask_check' (ffff82d0802d3370 != ffff82d080337b60) Duplicate symbol 'common.c#get_cpu_info' (ffff82d0802d22b0 != ffff82d080331590) Duplicate symbol 'common.c#get_cpu_info_from_stack' (ffff82d0802d31c0 != ffff82d0803374b0) Duplicate symbol 'common.c#pfn_to_pdx' (ffff82d0802d3270 != ffff82d080331e00) Duplicate symbol 'common.c#test_and_set_bit' (ffff82d0802d3360 != ffff82d080332250) Duplicate symbol 'common.c#variable_clear_bit' (ffff82d0802d2270 != ffff82d080337b50) Duplicate symbol 'compat.c#get_cpu_info' (ffff82d08026eab0 != ffff82d080200460) Duplicate symbol 'compat.c#get_cpu_info_from_stack' (ffff82d08026ebd0 != ffff82d080200f70) Duplicate symbol 'cpu_idle.c#get_cpu_info' (ffff82d0802ccb00 != ffff82d08035fcc0) [...]

For the time being disable live-patching when building with clang, since duplicate symbols will trigger a build failure because ENFORCE_UNIQUE_SYMBOLS is now also enabled by default in conjunction with live-patching.

Quoted 7 lines
Quoted 5 lines
Quoted 1 line

To be honest, as indicated before I'm inclined to nak this patch on the basis that a proper solution has been posted almost 3 weeks ago (and this was already v2).

What's that patch waiting on?

x86 and release acks.

Jan

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#4 lore raw
Quoted 30 lines · Jan Beulich
Quoted 29 lines · George Dunlap
Quoted 28 lines · Jan Beulich
Quoted 26 lines · Roger Pau Monne

Live-patching requires unique symbols, and sadly the clang build generates a lot of duplicate symbols:

Duplicate symbol 'asid.c#get_cpu_info' (ffff82d0803032c0 != ffff82d0802e0f50) Duplicate symbol 'asid.c#get_cpu_info_from_stack' (ffff82d0802e1080 != ffff82d0803032f0) Duplicate symbol 'ats.c#__list_add' (ffff82d080260a00 != ffff82d080267c70) Duplicate symbol 'boot.c#constant_test_bit' (ffff82d08040ea60 != ffff82d0804372f0) Duplicate symbol 'common.c#clear_bit' (ffff82d080332440 != ffff82d0802d33b0) Duplicate symbol 'common.c#constant_test_bit' (ffff82d080332340 != ffff82d0802d2220) Duplicate symbol 'common.c#cpumask_check' (ffff82d0802d3370 != ffff82d080337b60) Duplicate symbol 'common.c#get_cpu_info' (ffff82d0802d22b0 != ffff82d080331590) Duplicate symbol 'common.c#get_cpu_info_from_stack' (ffff82d0802d31c0 != ffff82d0803374b0) Duplicate symbol 'common.c#pfn_to_pdx' (ffff82d0802d3270 != ffff82d080331e00) Duplicate symbol 'common.c#test_and_set_bit' (ffff82d0802d3360 != ffff82d080332250) Duplicate symbol 'common.c#variable_clear_bit' (ffff82d0802d2270 != ffff82d080337b50) Duplicate symbol 'compat.c#get_cpu_info' (ffff82d08026eab0 != ffff82d080200460) Duplicate symbol 'compat.c#get_cpu_info_from_stack' (ffff82d08026ebd0 != ffff82d080200f70) Duplicate symbol 'cpu_idle.c#get_cpu_info' (ffff82d0802ccb00 != ffff82d08035fcc0) [...]

For the time being disable live-patching when building with clang, since duplicate symbols will trigger a build failure because ENFORCE_UNIQUE_SYMBOLS is now also enabled by default in conjunction with live-patching.

Quoted 7 lines
Quoted 5 lines
Quoted 3 lines

To be honest, as indicated before I'm inclined to nak this patch on the basis that a proper solution has been posted almost 3 weeks ago (and this was already v2).

What's that patch waiting on?

x86 and release acks.

I plan to release ack the patch in case the missing maintainer's acks are not coming in too late.

Juergen

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#5 lore raw
Quoted 50 lines · Jürgen Groß
Quoted 49 lines · Jan Beulich
Quoted 48 lines · George Dunlap
Quoted 46 lines · Jan Beulich
Quoted 35 lines · Roger Pau Monne

Live-patching requires unique symbols, and sadly the clang build generates a lot of duplicate symbols:

Duplicate symbol 'asid.c#get_cpu_info' (ffff82d0803032c0 != ffff82d0802e0f50)
Duplicate symbol 'asid.c#get_cpu_info_from_stack' (ffff82d0802e1080 != ffff82d0803032f0)
Duplicate symbol 'ats.c#__list_add' (ffff82d080260a00 != ffff82d080267c70)
Duplicate symbol 'boot.c#constant_test_bit' (ffff82d08040ea60 != ffff82d0804372f0)
Duplicate symbol 'common.c#clear_bit' (ffff82d080332440 != ffff82d0802d33b0)
Duplicate symbol 'common.c#constant_test_bit' (ffff82d080332340 != ffff82d0802d2220)
Duplicate symbol 'common.c#cpumask_check' (ffff82d0802d3370 != ffff82d080337b60)
Duplicate symbol 'common.c#get_cpu_info' (ffff82d0802d22b0 != ffff82d080331590)
Duplicate symbol 'common.c#get_cpu_info_from_stack' (ffff82d0802d31c0 != ffff82d0803374b0)
Duplicate symbol 'common.c#pfn_to_pdx' (ffff82d0802d3270 != ffff82d080331e00)
Duplicate symbol 'common.c#test_and_set_bit' (ffff82d0802d3360 != ffff82d080332250)
Duplicate symbol 'common.c#variable_clear_bit' (ffff82d0802d2270 != ffff82d080337b50)
Duplicate symbol 'compat.c#get_cpu_info' (ffff82d08026eab0 != ffff82d080200460)
Duplicate symbol 'compat.c#get_cpu_info_from_stack' (ffff82d08026ebd0 != ffff82d080200f70)
Duplicate symbol 'cpu_idle.c#get_cpu_info' (ffff82d0802ccb00 != ffff82d08035fcc0)
[...]

Quoted 6 lines

For the time being disable live-patching when building with clang, since duplicate symbols will trigger a build failure because ENFORCE_UNIQUE_SYMBOLS is now also enabled by default in conjunction with live-patching.

To be honest, as indicated before I'm inclined to nak this patch on the basis that a proper solution has been posted almost 3 weeks ago (and this was already v2).

Quoted 6 lines
Quoted 3 lines
Quoted 1 line

What's that patch waiting on?

x86 and release acks.

I plan to release ack the patch in case the missing maintainer's acks are not coming in too late.

I think Andy's objection was that there has been zero testing of livepatching on gcc. Maybe we can find someone to do a smoke-test.

-George

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#6 lore raw
Quoted 5 lines
Quoted 2 lines

I plan to release ack the patch in case the missing maintainer's acks are not coming in too late.

I think Andy's objection was that there has been zero testing of livepatching on gcc. Maybe we can find someone to do a smoke-test.

As in integrate livepatch-build tools in osstest smoke-tests? Because the livepatch test cases are in osstest, unless something went awry?

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#7 lore raw
Quoted 6 lines · Konrad Rzeszutek Wilk
Quoted 4 lines
Quoted 2 lines

I plan to release ack the patch in case the missing maintainer's acks are not coming in too late.

I think Andy's objection was that there has been zero testing of livepatching on gcc. Maybe we can find someone to do a smoke-test.

As in integrate livepatch-build tools in osstest smoke-tests? Because the livepatch test cases are in osstest, unless something went awry?

The sum total of livepatch testing in OSSTest is using the hand-coded ELF objects from the tests/ directory.

This is perhaps ok for the basic mechanism, but its not representative of actually building real livepatches using livepatch build tools.

~Andrew

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#8 lore raw
Quoted 5 lines · Andrew Cooper
Quoted 4 lines · Konrad Rzeszutek Wilk
Quoted 4 lines
Quoted 2 lines

I plan to release ack the patch in case the missing maintainer's acks are not coming in too late.

I think Andy's objection was that there has been zero testing of livepatching on gcc. Maybe we can find someone to do a smoke-test.

Quoted 8 lines
Quoted 2 lines

As in integrate livepatch-build tools in osstest smoke-tests? Because the livepatch test cases are in osstest, unless something went awry?

The sum total of livepatch testing in OSSTest is using the hand-coded ELF objects from the tests/ directory.

This is perhaps ok for the basic mechanism, but its not representative of actually building real livepatches using livepatch build tools.

True. But it tests the _hypervisor_ livepatch code.

I am thinking that this discussion about "oh, but livepatch-build tools don't work b/c" is well <shrug> sucks but should never block an release as the core livepatch functionality is OK.

Irrespective of that the testing of livepatch-build tools should be in osstest, granted nobody has taken a step in this - but is somebody signing up for it? [I can't, -ENOTIME]

Quoted 2 lines

~Andrew

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#9 lore raw
Quoted 16 lines · Konrad Rzeszutek Wilk
Quoted 11 lines · Andrew Cooper
Quoted 6 lines · Konrad Rzeszutek Wilk
Quoted 4 lines
Quoted 2 lines

I plan to release ack the patch in case the missing maintainer's acks are not coming in too late.

I think Andy's objection was that there has been zero testing of livepatching on gcc. Maybe we can find someone to do a smoke-test.

As in integrate livepatch-build tools in osstest smoke-tests? Because the livepatch test cases are in osstest, unless something went awry?

The sum total of livepatch testing in OSSTest is using the hand-coded ELF objects from the tests/ directory.

Quoted 8 lines
Quoted 2 lines

This is perhaps ok for the basic mechanism, but its not representative of actually building real livepatches using livepatch build tools.

True. But it tests the _hypervisor_ livepatch code.

I am thinking that this discussion about "oh, but livepatch-build tools don't work b/c" is well <shrug> sucks but should never block an release as the core livepatch functionality is OK.

I think a parallel is if Xen doesn’t build with a particular version of the compiler, or can’t build on a particular distro for some reason. We should certainly *try* to make things work with other projects, but if the issue is clearly with the other project, we shouldn’t have to block to wait for that other project to get things sorted out.

-George
_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#10 lore raw
Quoted 14 lines · George Dunlap
Quoted 14 lines · Konrad Rzeszutek Wilk
Quoted 10 lines · Andrew Cooper
Quoted 6 lines · Konrad Rzeszutek Wilk
Quoted 4 lines
Quoted 2 lines

I plan to release ack the patch in case the missing maintainer's acks are not coming in too late.

I think Andy's objection was that there has been zero testing of livepatching on gcc. Maybe we can find someone to do a smoke-test.

As in integrate livepatch-build tools in osstest smoke-tests? Because the livepatch test cases are in osstest, unless something went awry?

The sum total of livepatch testing in OSSTest is using the hand-coded ELF objects from the tests/ directory.

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

This is perhaps ok for the basic mechanism, but its not representative of actually building real livepatches using livepatch build tools.

True. But it tests the _hypervisor_ livepatch code.

I am thinking that this discussion about "oh, but livepatch-build tools don't work b/c" is well <shrug> sucks but should never block an release as the core livepatch functionality is OK.

I think a parallel is if Xen doesn’t build with a particular version of the compiler, or can’t build on a particular distro for some reason. We should certainly *try* to make things work with other projects, but if the issue is clearly with the other project, we shouldn’t have to block to wait for that other project to get things sorted out.

This isn't a valid comparison.

livepatch-build-tools is a concrete thing, built and maintained by us (the Xen community), explicitly for the purpose generating livepatches between two versions of Xen.  It lives at
https://xenbits.xen.org/gitweb/?p=livepatch-build-tools.git;a=summary on xenbits, just like xen.git.

It *should* be used in OSSTest, have a push gate, and block breaking changes either to Xen or to the tools themselves, before the breaking changes get accepted into master of either repo.

Otherwise, the support statement for livepatching needs to change to be "we don't bother testing the two parts of this supported feature together".

~Andrew

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel

#11 lore raw
Quoted 28 lines · Andrew Cooper
Quoted 14 lines · George Dunlap
Quoted 14 lines · Konrad Rzeszutek Wilk
Quoted 10 lines · Andrew Cooper
Quoted 6 lines · Konrad Rzeszutek Wilk
Quoted 4 lines
Quoted 2 lines

I plan to release ack the patch in case the missing maintainer's acks are not coming in too late.

I think Andy's objection was that there has been zero testing of livepatching on gcc. Maybe we can find someone to do a smoke-test.

As in integrate livepatch-build tools in osstest smoke-tests? Because the livepatch test cases are in osstest, unless something went awry?

The sum total of livepatch testing in OSSTest is using the hand-coded ELF objects from the tests/ directory.

Quoted 8 lines
Quoted 7 lines
Quoted 2 lines

This is perhaps ok for the basic mechanism, but its not representative of actually building real livepatches using livepatch build tools.

True. But it tests the _hypervisor_ livepatch code.

I am thinking that this discussion about "oh, but livepatch-build tools don't work b/c" is well <shrug> sucks but should never block an release as the core livepatch functionality is OK.

I think a parallel is if Xen doesn’t build with a particular version of the compiler, or can’t build on a particular distro for some reason. We should certainly *try* to make things work with other projects, but if the issue is clearly with the other project, we shouldn’t have to block to wait for that other project to get things sorted out.

Quoted 7 lines

This isn't a valid comparison.

livepatch-build-tools is a concrete thing, built and maintained by us
(the Xen community), explicitly for the purpose generating livepatches
between two versions of Xen. It lives at
https://xenbits.xen.org/gitweb/?p=livepatch-build-tools.git;a=summary <https://xenbits.xen.org/gitweb/?p=livepatch-build-tools.git;a=summary> on xenbits, just like xen.git.

First a couple of questions: I noticed that neither Ross to xen-devel is on this thread

I agree with Andy: we got away lucky so far, as there have been few changes to the live patch-build-tools

Quoted 3 lines

It *should* be used in OSSTest, have a push gate, and block breaking changes either to Xen or to the tools themselves, before the breaking changes get accepted into master of either repo.

Although I agree with you, we should not block 4.13 for it and do some manual testing for this release But we should have a plan in place for 4.14 to address this and maybe agree to block 4.14 if that has not happened

Lars

_______________________________________________
Xen-devel mailing list
[email protected]
https://lists.xenproject.org/mailman/listinfo/xen-devel