From: Tengda Wu <wutengda@huaweicloud.com>
To: Namhyung Kim <namhyung@kernel.org>,
james.clark@linaro.org, xueshuai@linux.alibaba.com,
Adrian Hunter <adrian.hunter@intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>,
leo.yan@linux.dev, Li Huafei <lihuafei1@huawei.com>,
Ian Rogers <irogers@google.com>,
Kim Phillips <kim.phillips@arm.com>,
Mark Rutland <mark.rutland@arm.com>,
Arnaldo Carvalho de Melo <acme@kernel.org>,
Ingo Molnar <mingo@redhat.com>, Bill Wendling <morbo@google.com>,
Nick Desaulniers <nick.desaulniers+lkml@gmail.com>,
Alexander Shishkin <alexander.shishkin@linux.intel.com>,
Zecheng Li <zli94@ncsu.edu>,
linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org,
llvm@lists.linux.dev, Tengda Wu <wutengda@huaweicloud.com>
Subject: [PATCH v5 02/26] perf llvm: Fix arm64 adrp instruction disassembly mismatch with objdump
Date: Tue, 8 Sep 2026 13:05:05 +0000 [thread overview]
Message-ID: <aca3c51ca091bdf0c4b49b0721fbbf78a05abcf1.1788872630.git.wutengda@huaweicloud.com> (raw)
In-Reply-To: <cover.1788872630.git.wutengda@huaweicloud.com>
The operands of 'adrp' instructions parsed by libllvm are currently
represented as raw immediates rather than the "address <symbol+offset>"
format used by objdump. This inconsistency causes arm64_mov__parse()
to fail when parsing these instructions during post-processing.
Example of the mismatch:
Current: adrp x18, 8014
Objdump: adrp x18, ffff800081f5f000 <this_cpu_vector>
Fix this by manually extracting the target address from the raw adrp
instruction via symbol_lookup_callback(). The address is then converted
to a specific symbol during symbol__disassemble_llvm() and formatted
to match objdump's output, ensuring compatibility with existing
parsers.
Signed-off-by: Tengda Wu <wutengda@huaweicloud.com>
---
tools/perf/util/llvm.c | 52 ++++++++++++++++++++++++++++++++++++++----
1 file changed, 47 insertions(+), 5 deletions(-)
diff --git a/tools/perf/util/llvm.c b/tools/perf/util/llvm.c
index a0deb742a733..a78845689aa1 100644
--- a/tools/perf/util/llvm.c
+++ b/tools/perf/util/llvm.c
@@ -9,8 +9,10 @@
#include "symbol.h"
#include <errno.h>
#include <fcntl.h>
+#include <string.h>
#include <unistd.h>
#include <linux/zalloc.h>
+#include <linux/bitops.h>
#ifdef HAVE_LIBLLVM_SUPPORT
#include "llvm-c-helpers.h"
@@ -108,6 +110,16 @@ symbol_lookup_callback(void *disinfo, uint64_t value,
storage->branch_addr = value;
else if (*ref_type == LLVMDisassembler_ReferenceType_In_PCrel_Load)
storage->pcrel_load_addr = value;
+ else if (*ref_type == LLVMDisassembler_ReferenceType_In_ARM64_ADRP) {
+ uint64_t adrp_imm;
+
+ /* immhi (bits 23:5) and immlo (bits 30:29) */
+ adrp_imm = ((value & 0x00ffffe0) >> 3) | ((value >> 29) & 0x3);
+ adrp_imm = sign_extend64(adrp_imm, 20);
+
+ /* Calculate the target page address */
+ storage->pcrel_load_addr = (address & ~0xfffULL) + (adrp_imm << 12);
+ }
*ref_type = LLVMDisassembler_ReferenceType_InOut_None;
return NULL;
}
@@ -228,12 +240,42 @@ int symbol__disassemble_llvm(const char *filename, struct symbol *sym,
}
}
if (storage.pcrel_load_addr != 0) {
- char *name = llvm_name_for_data(dso, filename,
+ char *name;
+
+ if (arch__is_arm64(args->arch)) {
+ /*
+ * For arm64, replace the immediate operand with
+ * the resolved address to match objdump's output.
+ *
+ * Example conversion:
+ * From: adrp x18, 8014
+ * To: adrp x18, ffff800081f5f000
+ */
+ char *s = strchr(disasm_buf, ',');
+
+ if (s == NULL)
+ goto err;
+
+ s++;
+ *s = '\0';
+ disasm_len = strlen(disasm_buf);
+ disasm_len += scnprintf(disasm_buf + disasm_len,
+ sizeof(disasm_buf) - disasm_len,
+ " %"PRIx64,
storage.pcrel_load_addr);
- disasm_len += scnprintf(disasm_buf + disasm_len,
- sizeof(disasm_buf) - disasm_len,
- " # %#"PRIx64,
- storage.pcrel_load_addr);
+ } else {
+ /*
+ * For other archs, append the resolved address
+ * as an inline comment.
+ */
+ disasm_len += scnprintf(disasm_buf + disasm_len,
+ sizeof(disasm_buf) - disasm_len,
+ " # %#"PRIx64,
+ storage.pcrel_load_addr);
+ }
+
+ /* Append the resolved symbol name if available */
+ name = llvm_name_for_data(dso, filename, storage.pcrel_load_addr);
if (name) {
disasm_len += scnprintf(disasm_buf + disasm_len,
sizeof(disasm_buf) -
--
2.34.1
next prev parent reply other threads:[~2026-09-08 13:05 UTC|newest]
Thread overview: 74+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-09-08 13:05 [PATCH v5 00/26] perf arm64: Support data type profiling Tengda Wu
2026-09-08 13:05 ` [PATCH v5 01/26] perf capstone: Symbolize address operands to match objdump on arm64 Tengda Wu
2026-09-08 13:18 ` sashiko-bot
2026-09-09 14:49 ` Namhyung Kim
2026-09-10 7:38 ` Tengda Wu
2026-09-08 13:05 ` Tengda Wu [this message]
2026-09-08 13:16 ` [PATCH v5 02/26] perf llvm: Fix arm64 adrp instruction disassembly mismatch with objdump sashiko-bot
2026-09-09 14:50 ` Namhyung Kim
2026-09-08 13:05 ` [PATCH v5 03/26] perf annotate-arm64: Generalize arm64_mov__parse to support more instructions Tengda Wu
2026-09-08 13:15 ` sashiko-bot
2026-09-09 14:52 ` Namhyung Kim
2026-09-08 13:05 ` [PATCH v5 04/26] perf annotate-arm64: Handle load and store instructions Tengda Wu
2026-09-08 13:15 ` sashiko-bot
2026-09-09 14:58 ` Namhyung Kim
2026-09-08 13:05 ` [PATCH v5 05/26] perf annotate: Normalize arch__dwarf_regnum() error return values Tengda Wu
2026-09-08 13:24 ` sashiko-bot
2026-09-08 17:57 ` Ian Rogers
2026-09-09 15:00 ` Namhyung Kim
2026-09-08 13:05 ` [PATCH v5 06/26] perf annotate: Introduce extract_op_location callback for arch-specific parsing Tengda Wu
2026-09-08 13:20 ` sashiko-bot
2026-09-09 15:01 ` Namhyung Kim
2026-09-08 13:05 ` [PATCH v5 07/26] perf dwarf-regs: Adapt get_dwarf_regnum() for arm64 Tengda Wu
2026-09-08 13:21 ` sashiko-bot
2026-09-09 15:03 ` Namhyung Kim
2026-09-08 13:05 ` [PATCH v5 08/26] perf annotate: Adapt arch__dwarf_regnum() " Tengda Wu
2026-09-08 13:15 ` sashiko-bot
2026-09-09 15:04 ` Namhyung Kim
2026-09-08 13:05 ` [PATCH v5 09/26] perf annotate-arm64: Implement extract_op_location() callback Tengda Wu
2026-09-08 13:18 ` sashiko-bot
2026-09-10 8:49 ` Tengda Wu
2026-09-08 13:05 ` [PATCH v5 10/26] perf annotate: Default to --itrace=i1i for data type profiling Tengda Wu
2026-09-08 13:15 ` sashiko-bot
2026-09-09 15:38 ` Adrian Hunter
2026-09-08 13:05 ` [PATCH v5 11/26] perf arm-spe: Set default synthesized event period to 1 Tengda Wu
2026-09-08 13:12 ` sashiko-bot
2026-09-08 13:05 ` [PATCH v5 12/26] perf annotate-data: Extract invalidate_reg_state() as a common helper Tengda Wu
2026-09-08 13:11 ` sashiko-bot
2026-09-08 13:05 ` [PATCH v5 13/26] perf annotate-arm64: Enable instruction tracking support Tengda Wu
2026-09-08 13:18 ` sashiko-bot
2026-09-10 9:20 ` Tengda Wu
2026-09-08 13:05 ` [PATCH v5 14/26] perf annotate-data: Add arch_get_reg_offset helper Tengda Wu
2026-09-08 13:22 ` sashiko-bot
2026-09-10 12:13 ` Tengda Wu
2026-09-08 13:05 ` [PATCH v5 15/26] perf annotate-arm64: Track return type after call instructions Tengda Wu
2026-09-08 13:14 ` sashiko-bot
2026-09-08 13:05 ` [PATCH v5 16/26] perf annotate-arm64: Support load instruction tracking Tengda Wu
2026-09-08 13:23 ` sashiko-bot
2026-09-10 13:19 ` Tengda Wu
2026-09-10 13:28 ` Tengda Wu
2026-09-08 13:05 ` [PATCH v5 17/26] perf annotate-arm64: Support store " Tengda Wu
2026-09-08 13:22 ` sashiko-bot
2026-09-08 13:05 ` [PATCH v5 18/26] perf annotate-data: Expand type_state_reg imm_value to u64 Tengda Wu
2026-09-08 13:26 ` sashiko-bot
2026-09-08 13:05 ` [PATCH v5 19/26] perf annotate-data: Track imm_value for stack variables Tengda Wu
2026-09-08 13:29 ` sashiko-bot
2026-09-08 13:05 ` [PATCH v5 20/26] perf annotate-x86: Delete stale stack state on store of untracked register Tengda Wu
2026-09-08 13:20 ` sashiko-bot
2026-09-08 13:05 ` [PATCH v5 21/26] perf annotate-arm64: Support stack variable tracking Tengda Wu
2026-09-08 13:36 ` sashiko-bot
2026-09-11 1:41 ` Tengda Wu
2026-09-08 13:05 ` [PATCH v5 22/26] perf annotate-arm64: Support 'mov' instruction tracking Tengda Wu
2026-09-08 13:27 ` sashiko-bot
2026-09-11 2:22 ` Tengda Wu
2026-09-08 13:05 ` [PATCH v5 23/26] perf annotate-arm64: Support 'add' " Tengda Wu
2026-09-08 13:27 ` sashiko-bot
2026-09-11 2:29 ` Tengda Wu
2026-09-08 13:06 ` [PATCH v5 24/26] perf annotate-arm64: Support 'adrp' instruction to track global variables Tengda Wu
2026-09-08 13:26 ` sashiko-bot
2026-09-08 13:06 ` [PATCH v5 25/26] perf annotate-arm64: Support per-cpu variable access tracking Tengda Wu
2026-09-08 13:30 ` sashiko-bot
2026-09-11 10:15 ` Tengda Wu
2026-09-08 13:06 ` [PATCH v5 26/26] perf annotate-arm64: Support 'mrs' instruction to track 'current' pointer Tengda Wu
2026-09-08 13:31 ` sashiko-bot
-- strict thread matches above, loose matches on Subject: below --
2026-09-08 13:00 [PATCH v5 00/26] perf arm64: Support data type profiling Tengda Wu
2026-09-08 13:00 ` [PATCH v5 02/26] perf llvm: Fix arm64 adrp instruction disassembly mismatch with objdump Tengda Wu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=aca3c51ca091bdf0c4b49b0721fbbf78a05abcf1.1788872630.git.wutengda@huaweicloud.com \
--to=wutengda@huaweicloud.com \
--cc=acme@kernel.org \
--cc=adrian.hunter@intel.com \
--cc=alexander.shishkin@linux.intel.com \
--cc=irogers@google.com \
--cc=james.clark@linaro.org \
--cc=kim.phillips@arm.com \
--cc=leo.yan@linux.dev \
--cc=lihuafei1@huawei.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-perf-users@vger.kernel.org \
--cc=llvm@lists.linux.dev \
--cc=mark.rutland@arm.com \
--cc=mingo@redhat.com \
--cc=morbo@google.com \
--cc=namhyung@kernel.org \
--cc=nick.desaulniers+lkml@gmail.com \
--cc=peterz@infradead.org \
--cc=xueshuai@linux.alibaba.com \
--cc=zli94@ncsu.edu \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox