Mailing List Archive: [PATCH 0/7] xen: Switch to using -Og for debug builds

[PATCH 0/7] xen: Switch to using -Og for debug builds

Apr 19, 2021, 7:01 AM

Post #1 of 4 (2 views)

As with the toolstack side, we ought to use -Og for debug builds.

All fixes are trivial. The first 3 are understandable, given reduced
optimisations. The next 3 are, AFAICT, bogus diagnostics.

Andrew Cooper (7):
xen/arm: Make make_cpus_node() compile at -Og
x86/shim: Fix compilation at -Og
x86/sysctl: Make arch_do_sysctl() compile at -Og
x86/irq: Make create_irq() compile at -Og
xen/efi: Make efi_start() compile at -Og
x86/shadow: Make _shadow_prealloc() compile at -Og
xen: Use -Og for debug builds when available

xen/Makefile | 4 +++-
xen/arch/arm/domain_build.c | 2 +-
xen/arch/x86/irq.c | 2 +-
xen/arch/x86/mm/shadow/common.c | 2 +-
xen/arch/x86/pv/shim.c | 6 +++---
xen/arch/x86/sysctl.c | 4 ++--
xen/common/efi/boot.c | 2 +-
7 files changed, 12 insertions(+), 10 deletions(-)

--
2.11.0

Re: [PATCH 0/7] xen: Switch to using -Og for debug builds [ In reply to ]

jbeulich at suse

Apr 19, 2021, 8:45 AM

Post #2 of 4 (2 views)

Permalink

On 19.04.2021 16:01, Andrew Cooper wrote:
> As with the toolstack side, we ought to use -Og for debug builds.
>
> All fixes are trivial. The first 3 are understandable, given reduced
> optimisations. The next 3 are, AFAICT, bogus diagnostics.
>
> Andrew Cooper (7):
> xen/arm: Make make_cpus_node() compile at -Og
> x86/shim: Fix compilation at -Og
> x86/sysctl: Make arch_do_sysctl() compile at -Og
> x86/irq: Make create_irq() compile at -Og
> xen/efi: Make efi_start() compile at -Og
> x86/shadow: Make _shadow_prealloc() compile at -Og
> xen: Use -Og for debug builds when available

Acked-by: Jan Beulich <jbeulich@suse.com>

I'd like to ask though that at least for the bogus warnings you
amend the commit messages with the gcc version these were observed
with. Perhaps even those seemingly bogus initializers would
benefit from a very brief comment (or else there's a fair chance
of them getting removed again at some point).

Jan

Re: [PATCH 0/7] xen: Switch to using -Og for debug builds [ In reply to ]

andrew.cooper3 at citrix

Apr 21, 2021, 2:31 AM

Post #3 of 4 (1 views)

Permalink

On 19/04/2021 16:45, Jan Beulich wrote:
> On 19.04.2021 16:01, Andrew Cooper wrote:
>> As with the toolstack side, we ought to use -Og for debug builds.
>>
>> All fixes are trivial. The first 3 are understandable, given reduced
>> optimisations. The next 3 are, AFAICT, bogus diagnostics.
>>
>> Andrew Cooper (7):
>> xen/arm: Make make_cpus_node() compile at -Og
>> x86/shim: Fix compilation at -Og
>> x86/sysctl: Make arch_do_sysctl() compile at -Og
>> x86/irq: Make create_irq() compile at -Og
>> xen/efi: Make efi_start() compile at -Og
>> x86/shadow: Make _shadow_prealloc() compile at -Og
>> xen: Use -Og for debug builds when available
> Acked-by: Jan Beulich <jbeulich@suse.com>
>
> I'd like to ask though that at least for the bogus warnings you
> amend the commit messages with the gcc version these were observed
> with. Perhaps even those seemingly bogus initializers would
> benefit from a very brief comment (or else there's a fair chance
> of them getting removed again at some point).

I'm afraid I don't have that information easily to hand, but all issues
were found by distro-provided compilers included in our Gitlab
infrastructure.

~Andrew

Re: [PATCH 0/7] xen: Switch to using -Og for debug builds [ In reply to ]

jbeulich at suse

Apr 21, 2021, 2:51 AM

Post #4 of 4 (1 views)

Permalink

On 21.04.2021 11:31, Andrew Cooper wrote:
> On 19/04/2021 16:45, Jan Beulich wrote:
>> On 19.04.2021 16:01, Andrew Cooper wrote:
>>> As with the toolstack side, we ought to use -Og for debug builds.
>>>
>>> All fixes are trivial. The first 3 are understandable, given reduced
>>> optimisations. The next 3 are, AFAICT, bogus diagnostics.
>>>
>>> Andrew Cooper (7):
>>> xen/arm: Make make_cpus_node() compile at -Og
>>> x86/shim: Fix compilation at -Og
>>> x86/sysctl: Make arch_do_sysctl() compile at -Og
>>> x86/irq: Make create_irq() compile at -Og
>>> xen/efi: Make efi_start() compile at -Og
>>> x86/shadow: Make _shadow_prealloc() compile at -Og
>>> xen: Use -Og for debug builds when available
>> Acked-by: Jan Beulich <jbeulich@suse.com>
>>
>> I'd like to ask though that at least for the bogus warnings you
>> amend the commit messages with the gcc version these were observed
>> with. Perhaps even those seemingly bogus initializers would
>> benefit from a very brief comment (or else there's a fair chance
>> of them getting removed again at some point).
>
> I'm afraid I don't have that information easily to hand, but all issues
> were found by distro-provided compilers included in our Gitlab
> infrastructure.

Well, okay, then we'll need to live with said risk. As an aside I'd
like to remind you though that on past occasions of working around
compiler oddities, I was asked as well to provide compiler version
information ...

Jan