On 2023-03-27 19:51:34, Nicholas Piggin wrote:

On Mon Mar 27, 2023 at 7:34 PM AEST, Kautuk Consul wrote:

quoted

On 2023-03-27 14:58:03, Kautuk Consul wrote:

quoted

On 2023-03-27 19:19:37, Nicholas Piggin wrote:

quoted

On Thu Mar 16, 2023 at 3:10 PM AEST, Kautuk Consul wrote:

quoted

kvmppc_hv_entry isn't called from anywhere other than
book3s_hv_rmhandlers.S itself. Removing .global scope for
this function and annotating it with SYM_INNER_LABEL.

Signed-off-by: Kautuk Consul <redacted>
---
 arch/powerpc/kvm/book3s_hv_rmhandlers.S | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/arch/powerpc/kvm/book3s_hv_rmhandlers.S b/arch/powerpc/kvm/book3s_hv_rmhandlers.S
index acf80915f406..b81ba4ee0521 100644
--- a/arch/powerpc/kvm/book3s_hv_rmhandlers.S
+++ b/arch/powerpc/kvm/book3s_hv_rmhandlers.S

@@ -502,8 +502,7 @@ END_FTR_SECTION_IFSET(CPU_FTR_ARCH_207S)
  *                                                                            *
  *****************************************************************************/
 
-.global kvmppc_hv_entry

I think this is okay.

quoted

-kvmppc_hv_entry:
+SYM_INNER_LABEL(kvmppc_hv_entry, SYM_L_LOCAL)

The documentation for SYM_INNER_LABEL says it for labels inside a SYM
function block, is that a problem? This is a function but doesn't have
C calling convention, so asm annotation docs say that it should use
SYM_CODE_START_LOCAL?

That is correct. Will create a v4 patch for this and send it.

But using SYM_CODE_START_LOCAL again causes a warning in the build
(which we were trying to avoid):
arch/powerpc/kvm/book3s_hv_rmhandlers.o: warning: objtool: .text+0x48: unannotated intra-function call

Are you using SYM_FUNC_END as well? Looks like you need that to
annotate the type properly. It should be the same as SYM_INNER_LABEL
in the end AFAIKS.

What about SYM_CODE_START_LOCAL and SYM_CODE_END ?
This seems to work fine for me without any build warnings from objtool.

quoted

quoted

quoted

BTW. why don't our _GLOBAL() macros use these SYM annotations? I haven't
really looked into them.

Not sure. Was mostly just concentrating on the kvmppc_hv_entry code.

Looks like it's because we have a .type @function annotation in those
already. Not sure if we should end up converting all that over to use
the SYM annotations or if it's okay to leave it as is.

Thanks,
Nick