Re: Build Warnings at arch/powerpc/

From: Christophe Leroy
Date: Tue Mar 04 2025 - 08:50:08 EST




Le 04/03/2025 à 14:03, Venkat Rao Bagalkote a écrit :
[Vous ne recevez pas souvent de courriers de venkat88@xxxxxxxxxxxxx. Découvrez pourquoi ceci est important à https://aka.ms/ LearnAboutSenderIdentification ]

On 04/03/25 6:08 pm, Madhavan Srinivasan wrote:

On 3/4/25 4:58 PM, Madhavan Srinivasan wrote:

On 3/4/25 2:26 PM, Christophe Leroy wrote:

Le 04/03/2025 à 07:13, Madhavan Srinivasan a écrit :

On 3/4/25 10:42 AM, Venkat Rao Bagalkote wrote:
Greetings!!


Observing build warnings with linux-next and powerpc repo's. Issue is currently not seen on mainline yet.

PPC Repo: https://eur01.safelinks.protection.outlook.com/? url=https%3A%2F%2Fgit.kernel.org%2Fpub%2Fscm%2Flinux%2Fkernel%2Fgit%2Fpowerpc%2Flinux.git&data=05%7C02%7Cchristophe.leroy%40csgroup.eu%7C6e17cc771a204b2998b508dd5b1cf2cf%7C8b87af7d86474dc78df45f69a2011bb5%7C0%7C0%7C638766902127463526%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=f0ubC0BiY%2Fw2XLfXcX955JKhJ%2BRkUmTUVO4fV%2F%2F4v2Y%3D&reserved=0 merge branch

PPC Kernel Version: 6.14.0-rc4-g1304f486dbf1
next Repo: https://eur01.safelinks.protection.outlook.com/? url=https%3A%2F%2Fgit.kernel.org%2Fpub%2Fscm%2Flinux%2Fkernel%2Fgit%2Fnext%2Flinux-next.git&data=05%7C02%7Cchristophe.leroy%40csgroup.eu%7C6e17cc771a204b2998b508dd5b1cf2cf%7C8b87af7d86474dc78df45f69a2011bb5%7C0%7C0%7C638766902127477000%7CUnknown%7CTWFpbGZsb3d8eyJFbXB0eU1hcGkiOnRydWUsIlYiOiIwLjAuMDAwMCIsIlAiOiJXaW4zMiIsIkFOIjoiTWFpbCIsIldUIjoyfQ%3D%3D%7C0%7C%7C%7C&sdata=GSDtRBQ35owCeEpjMMCNiJw3iizdrUPQcHznop2BLeQ%3D&reserved=0 master branch

next Kernel Version: 6.14.0-rc5-next-20250303


On linux-next kernel issue got introduced b/w next-20250227 and next-20250303


Build Warnings:

arch/powerpc/kvm/book3s_hv_rmhandlers.o: warning: objtool: .text+0xe84: intra_function_call not a direct call
arch/powerpc/crypto/ghashp8-ppc.o: warning: objtool: .text+0x22c: unannotated intra-function call
arch/powerpc/kernel/switch.o: warning: objtool: .text+0x4: intra_function_call not a direct call


Can you please specific the compiler and compiler version you found this issue with

Can you also tell which defconfig you are using or provide your .config

It might also be helpfull if you can provide a disassembly of the three file.o around the warned address.
I could recreate the issue with gcc 11.4.1 20231218 with today's linux-next (but could not recreate with gcc 14 or gcc 11.3.0)

(20d5c66e1810 (HEAD -> master, tag: next-20250304, origin/master, origin/HEAD) Add linux-next specific files for 20250304)

warning for one of the switch.S file :

   CC      arch/powerpc/kernel/syscalls.o
   AS      arch/powerpc/kernel/switch.o
arch/powerpc/kernel/switch.o: warning: objtool: .text+0x4: intra_function_call not a direct call
I guess this is becos, for bl .+4, we recently added in the arch_decode_instruction (decode.c) to set the type as INSN_OTHER

         case 18: /* b[l][a] */
                 if (ins == 0x48000005)  /* bl .+4 */
                         typ = INSN_OTHER;

Which I think is the issue here, changing it to INSN_CALL from INSN_OTHER fixes the warning

diff --git a/tools/objtool/arch/powerpc/decode.c b/tools/objtool/arch/ powerpc/decode.c
index 26d5050424a9..ffd63a61a585 100644
--- a/tools/objtool/arch/powerpc/decode.c
+++ b/tools/objtool/arch/powerpc/decode.c
@@ -56,7 +56,7 @@ int arch_decode_instruction(struct objtool_file *file, const struct section *sec
         switch (opcode) {
         case 18: /* b[l][a] */
                 if (ins == 0x48000005)  /* bl .+4 */
-                       typ = INSN_OTHER;
+                       typ = INSN_CALL;
                 else if (ins & 1)       /* bl[a] */
                         typ = INSN_CALL;
                 else            /* b[a] */


Maddy

Maddy,

I changed the code manually and gave it a try. The Proposed fix,
partially fixes the issue. It gets rid of two of the warnings, but below
warning still persists.

It fixes the issue for you but will reintroduce the issue with clang.

The real fix is to remove the ANNOTATE_INTRA_FUNCTION_CALL in:

arch/powerpc/kernel/switch.S:42: ANNOTATE_INTRA_FUNCTION_CALL
arch/powerpc/kvm/book3s_hv_rmhandlers.S:1527: ANNOTATE_INTRA_FUNCTION_CALL
arch/powerpc/kvm/book3s_hv_rmhandlers.S:1534: ANNOTATE_INTRA_FUNCTION_CALL

Can you give it a try ?

For the last one, can you provide an assembly dump ? You get it with "objtool -dr arch/powerpc/crypto/ghashp8-ppc.o"

Also can you tell which defconfig you use or provide your .config if not standard one ?

Christophe



arch/powerpc/crypto/ghashp8-ppc.o: warning: objtool: .text+0x22c:
unannotated intra-function call

   CC      arch/powerpc/kernel/irq.o
   CC      arch/powerpc/kernel/align.o
   CC      arch/powerpc/kernel/signal_64.o

Objdump of switch.o:
arch/powerpc/kernel/switch.o:     file format elf64-powerpcle

Disassembly of section .text:

0000000000000000 <flush_branch_caches>:
        0:   a6 02 28 7d     mflr    r9
        4:   05 00 00 48     bl      8 <flush_branch_caches+0x8>
        8:   05 00 00 48     bl      c <flush_branch_caches+0xc>
        c:   05 00 00 48     bl      10 <flush_branch_caches+0x10>
       10:   05 00 00 48     bl      14 <flush_branch_caches+0x14>
       14:   05 00 00 48     bl      18 <flush_branch_caches+0x18>
       18:   05 00 00 48     bl      1c <flush_branch_caches+0x1c>
       1c:   05 00 00 48     bl      20 <flush_branch_caches+0x20>
       20:   05 00 00 48     bl      24 <flush_branch_caches+0x24>
       24:   05 00 00 48     bl      28 <flush_branch_caches+0x28>
       28:   05 00 00 48     bl      2c <flush_branch_caches+0x2c>


arch/powerpc/kernel/switch.S failing src section:

.balign 32
.global flush_branch_caches
flush_branch_caches:
         /* Save LR into r9 */
         mflr    r9

         // Flush the link stack
         .rept 64
         ANNOTATE_INTRA_FUNCTION_CALL
         bl      .+4
         .endr
         b       1f
         nops    6

Maddy


Christophe


Regards,

Venkat.