Hi Donglin, First bad commit (maybe != root cause): tree: https://git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git master head: 8d2be868b42c08290509c60515865f4de24ea704 commit: 3646970322464c21e69dcb9a2e37d461c5834bf5 [11225/12162] arm64: ftrace: Enable HAVE_FUNCTION_GRAPH_RETVAL config: arm64-randconfig-r015-20230622 (https://download.01.org/0day-ci/archive/20230624/202306240021.qNRc4iHW-lkp@xxxxxxxxx/config) compiler: aarch64-linux-gcc (GCC) 12.3.0 reproduce: (https://download.01.org/0day-ci/archive/20230624/202306240021.qNRc4iHW-lkp@xxxxxxxxx/reproduce) If you fix the issue in a separate patch/commit (i.e. not just a new version of the same patch/commit), kindly add following tags | Reported-by: kernel test robot <lkp@xxxxxxxxx> | Closes: https://lore.kernel.org/oe-kbuild-all/202306240021.qNRc4iHW-lkp@xxxxxxxxx/ All errors (new ones prefixed by >>): arch/arm64/kernel/entry-ftrace.S: Assembler messages: >> arch/arm64/kernel/entry-ftrace.S:59: Error: undefined symbol FTRACE_OPS_DIRECT_CALL used as an immediate value vim +59 arch/arm64/kernel/entry-ftrace.S 819e50e25d0ce8 AKASHI Takahiro 2014-04-30 15 26299b3f6ba26b Mark Rutland 2022-11-03 16 #ifdef CONFIG_DYNAMIC_FTRACE_WITH_ARGS 3b23e4991fb66f Torsten Duwe 2019-02-08 17 /* 3b23e4991fb66f Torsten Duwe 2019-02-08 18 * Due to -fpatchable-function-entry=2, the compiler has placed two NOPs before 3b23e4991fb66f Torsten Duwe 2019-02-08 19 * the regular function prologue. For an enabled callsite, ftrace_init_nop() and 3b23e4991fb66f Torsten Duwe 2019-02-08 20 * ftrace_make_call() have patched those NOPs to: 3b23e4991fb66f Torsten Duwe 2019-02-08 21 * 3b23e4991fb66f Torsten Duwe 2019-02-08 22 * MOV X9, LR 26299b3f6ba26b Mark Rutland 2022-11-03 23 * BL ftrace_caller 3b23e4991fb66f Torsten Duwe 2019-02-08 24 * 258c3d628fe9e7 Will Deacon 2020-05-18 25 * Each instrumented function follows the AAPCS, so here x0-x8 and x18-x30 are 258c3d628fe9e7 Will Deacon 2020-05-18 26 * live (x18 holds the Shadow Call Stack pointer), and x9-x17 are safe to 258c3d628fe9e7 Will Deacon 2020-05-18 27 * clobber. 3b23e4991fb66f Torsten Duwe 2019-02-08 28 * 26299b3f6ba26b Mark Rutland 2022-11-03 29 * We save the callsite's context into a struct ftrace_regs before invoking any 26299b3f6ba26b Mark Rutland 2022-11-03 30 * ftrace callbacks. So that we can get a sensible backtrace, we create frame 26299b3f6ba26b Mark Rutland 2022-11-03 31 * records for the callsite and the ftrace entry assembly. This is not 26299b3f6ba26b Mark Rutland 2022-11-03 32 * sufficient for reliable stacktrace: until we create the callsite stack 26299b3f6ba26b Mark Rutland 2022-11-03 33 * record, its caller is missing from the LR and existing chain of frame 26299b3f6ba26b Mark Rutland 2022-11-03 34 * records. 3b23e4991fb66f Torsten Duwe 2019-02-08 35 */ 26299b3f6ba26b Mark Rutland 2022-11-03 36 SYM_CODE_START(ftrace_caller) 26299b3f6ba26b Mark Rutland 2022-11-03 37 bti c 3b23e4991fb66f Torsten Duwe 2019-02-08 38 2aa6ac03516d07 Florent Revest 2023-04-05 39 #ifdef CONFIG_DYNAMIC_FTRACE_WITH_CALL_OPS 2aa6ac03516d07 Florent Revest 2023-04-05 40 /* 2aa6ac03516d07 Florent Revest 2023-04-05 41 * The literal pointer to the ops is at an 8-byte aligned boundary 2aa6ac03516d07 Florent Revest 2023-04-05 42 * which is either 12 or 16 bytes before the BL instruction in the call 2aa6ac03516d07 Florent Revest 2023-04-05 43 * site. See ftrace_call_adjust() for details. 2aa6ac03516d07 Florent Revest 2023-04-05 44 * 2aa6ac03516d07 Florent Revest 2023-04-05 45 * Therefore here the LR points at `literal + 16` or `literal + 20`, 2aa6ac03516d07 Florent Revest 2023-04-05 46 * and we can find the address of the literal in either case by 2aa6ac03516d07 Florent Revest 2023-04-05 47 * aligning to an 8-byte boundary and subtracting 16. We do the 2aa6ac03516d07 Florent Revest 2023-04-05 48 * alignment first as this allows us to fold the subtraction into the 2aa6ac03516d07 Florent Revest 2023-04-05 49 * LDR. 2aa6ac03516d07 Florent Revest 2023-04-05 50 */ 2aa6ac03516d07 Florent Revest 2023-04-05 51 bic x11, x30, 0x7 2aa6ac03516d07 Florent Revest 2023-04-05 52 ldr x11, [x11, #-(4 * AARCH64_INSN_SIZE)] // op 2aa6ac03516d07 Florent Revest 2023-04-05 53 2aa6ac03516d07 Florent Revest 2023-04-05 54 #ifdef CONFIG_DYNAMIC_FTRACE_WITH_DIRECT_CALLS 2aa6ac03516d07 Florent Revest 2023-04-05 55 /* 2aa6ac03516d07 Florent Revest 2023-04-05 56 * If the op has a direct call, handle it immediately without 2aa6ac03516d07 Florent Revest 2023-04-05 57 * saving/restoring registers. 2aa6ac03516d07 Florent Revest 2023-04-05 58 */ 2aa6ac03516d07 Florent Revest 2023-04-05 @59 ldr x17, [x11, #FTRACE_OPS_DIRECT_CALL] // op->direct_call 2aa6ac03516d07 Florent Revest 2023-04-05 60 cbnz x17, ftrace_caller_direct 2aa6ac03516d07 Florent Revest 2023-04-05 61 #endif 2aa6ac03516d07 Florent Revest 2023-04-05 62 #endif 2aa6ac03516d07 Florent Revest 2023-04-05 63 26299b3f6ba26b Mark Rutland 2022-11-03 64 /* Save original SP */ 26299b3f6ba26b Mark Rutland 2022-11-03 65 mov x10, sp 3b23e4991fb66f Torsten Duwe 2019-02-08 66 26299b3f6ba26b Mark Rutland 2022-11-03 67 /* Make room for ftrace regs, plus two frame records */ 26299b3f6ba26b Mark Rutland 2022-11-03 68 sub sp, sp, #(FREGS_SIZE + 32) 3b23e4991fb66f Torsten Duwe 2019-02-08 69 26299b3f6ba26b Mark Rutland 2022-11-03 70 /* Save function arguments */ 26299b3f6ba26b Mark Rutland 2022-11-03 71 stp x0, x1, [sp, #FREGS_X0] 26299b3f6ba26b Mark Rutland 2022-11-03 72 stp x2, x3, [sp, #FREGS_X2] 26299b3f6ba26b Mark Rutland 2022-11-03 73 stp x4, x5, [sp, #FREGS_X4] 26299b3f6ba26b Mark Rutland 2022-11-03 74 stp x6, x7, [sp, #FREGS_X6] 26299b3f6ba26b Mark Rutland 2022-11-03 75 str x8, [sp, #FREGS_X8] 3b23e4991fb66f Torsten Duwe 2019-02-08 76 :::::: The code at line 59 was first introduced by commit :::::: 2aa6ac03516d078cf0c35aaa273b5cd11ea9734c arm64: ftrace: Add direct call support :::::: TO: Florent Revest <revest@xxxxxxxxxxxx> :::::: CC: Will Deacon <will@xxxxxxxxxx> -- 0-DAY CI Kernel Test Service https://github.com/intel/lkp-tests/wiki
