On 10.04.2024 12:24, Edgar E. Iglesias wrote:
> On Wed, Apr 10, 2024 at 12:21 PM Andrew Cooper <andrew.coop...@citrix.com>
> wrote:
>
>> On 10/04/2024 10:19 am, Edgar E. Iglesias wrote:
>>> From: "Edgar E. Iglesias" <edgar.igles...@amd.com>
>>>
>>> Use the generic xen/linkage.h macros when annotating symbols.
>>>
>>> Signed-off-by: Edgar E. Iglesias <edgar.igles...@amd.com>
>>> ---
>>> xen/arch/arm/arm64/entry.S | 12 ++++++------
>>> 1 file changed, 6 insertions(+), 6 deletions(-)
>>>
>>> diff --git a/xen/arch/arm/arm64/entry.S b/xen/arch/arm/arm64/entry.S
>>> index f963c923bb..6188dd2416 100644
>>> --- a/xen/arch/arm/arm64/entry.S
>>> +++ b/xen/arch/arm/arm64/entry.S
>>> @@ -480,9 +480,9 @@ guest_fiq_invalid_compat:
>>> guest_error_compat:
>>> guest_vector compat=1, iflags=IFLAGS__AI_, trap=guest_serror
>>>
>>> -ENTRY(return_to_new_vcpu32)
>>> +FUNC(return_to_new_vcpu32)
>>> exit hyp=0, compat=1
>>
>> In the new world, you want an END() too, which sets the size of the symbol.
>>
>> A good cross-check of this annotation stuff is:
>>
>> readelf -Wa xen-syms | grep return_to_new_vcpu32
>>
>> which in this case will tell you that the symbol called
>> return_to_new_vcpu32 still has a size of 0.
>
> Patch 2/2 adds the END, I should probably have squashed them into one...
Only partly afaics: return_to_new_vcpu{32,64} are still left without. And
yes, preferably the adjustments to the start annotation for a symbol
would come with an END() addition right away.
Jan
