Skip to contentSkip to conversation

riscv: uprobes: Add missing fence.i after building the XOL buffer

PATCH In review Björn Töpel 3 participants 4 messages Apr 17, 2025 – Apr 19, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 riscv: uprobes: Add missing fence.i after building the XOL buffer

From: Björn Töpel <[email protected]>

The XOL (execute out-of-line) buffer is used to single-step the replaced instruction(s) for uprobes. The RISC-V port was missing a proper fence.i (i$ flushing) after constructing the XOL buffer, which can result in incorrect execution of stale/broken instructions.

This was found running the BPF selftests "test_progs:
uprobe_autoattach, attach_probe" on the Spacemit K1/X60, where the uprobes tests randomly blew up.

Fixes 74784081aac8 ("riscv: Add uprobes supported")
---
 arch/riscv/kernel/probes/uprobes.c | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)
arch/riscv/kernel/probes/uprobes.c +1 -8
index 4b3dc8beaf77..4faef92dd771 100644
@@ -176,13 +176,6 @@ void arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr,
 		*(uprobe_opcode_t *)dst = __BUG_INSN_32;
 	}
 
+	flush_icache_range((unsigned long)dst, (unsigned long)dst + len);
 	kunmap_atomic(kaddr);
-
-	/*
-	 * We probably need flush_icache_user_page() but it needs vma.
-	 * This should work on most of architectures by default. If
-	 * architecture needs to do something different it can define
-	 * its own version of the function.
-	 */
-	flush_dcache_page(page);
 }

base-commit: 1a1d569a75f3ab2923cb62daf356d102e4df2b86

Signature · 1 line
2.45.2
Reviewed-by on v1 1/1
#1 lore raw
Quoted 11 lines · Björn Töpel

From: Björn Töpel <[email protected]>

The XOL (execute out-of-line) buffer is used to single-step the replaced instruction(s) for uprobes. The RISC-V port was missing a proper fence.i (i$ flushing) after constructing the XOL buffer, which can result in incorrect execution of stale/broken instructions.

This was found running the BPF selftests "test_progs:
uprobe_autoattach, attach_probe" on the Spacemit K1/X60, where the uprobes tests randomly blew up.

Good catch, Thx!

Quoted 31 lines
Fixes 74784081aac8 ("riscv: Add uprobes supported")
---
 arch/riscv/kernel/probes/uprobes.c | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)
arch/riscv/kernel/probes/uprobes.c +1 -8
index 4b3dc8beaf77..4faef92dd771 100644
@@ -176,13 +176,6 @@ void arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr,
                *(uprobe_opcode_t *)dst = __BUG_INSN_32;
        }
 
+       flush_icache_range((unsigned long)dst, (unsigned long)dst + len);
        kunmap_atomic(kaddr);
-
-       /*
-        * We probably need flush_icache_user_page() but it needs vma.
-        * This should work on most of architectures by default. If
-        * architecture needs to do something different it can define
-        * its own version of the function.
-        */
-       flush_dcache_page(page);
 }

base-commit: 1a1d569a75f3ab2923cb62daf356d102e4df2b86 --
2.45.2

Signature · 2 lines
Best Regards
 Guo Ren
#2 lore raw
In reply to Björn Töpel

Hi Björn,

Quoted 22 lines · Björn Töpel

From: Björn Töpel <[email protected]>

The XOL (execute out-of-line) buffer is used to single-step the replaced instruction(s) for uprobes. The RISC-V port was missing a proper fence.i (i$ flushing) after constructing the XOL buffer, which can result in incorrect execution of stale/broken instructions.

This was found running the BPF selftests "test_progs:
uprobe_autoattach, attach_probe" on the Spacemit K1/X60, where the uprobes tests randomly blew up.

Fixes 74784081aac8 ("riscv: Add uprobes supported")
---
 arch/riscv/kernel/probes/uprobes.c | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)
arch/riscv/kernel/probes/uprobes.c +0 -0
index 4b3dc8beaf77..4faef92dd771 100644

arch/riscv/kernel/probes/uprobes.c void arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr, · line 179

Quoted 4 lines
arch/riscv/kernel/probes/uprobes.c +1 -0
@@ -176,13 +176,6 @@ void arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr,
 		*(uprobe_opcode_t *)dst = __BUG_INSN_32;
 	}
 
+	flush_icache_range((unsigned long)dst, (unsigned long)dst + len);

This works because flush_icache_range currently ignores the range, but semantically is not quite right, because of the line just above the context that increments dst. If the range was respected, this would only flush the ebreak, not the preceding single-stepped instruction.

Regards,
Samuel

Quoted 12 lines
arch/riscv/kernel/probes/uprobes.c +0 -8
 	kunmap_atomic(kaddr);
-
-	/*
-	 * We probably need flush_icache_user_page() but it needs vma.
-	 * This should work on most of architectures by default. If
-	 * architecture needs to do something different it can define
-	 * its own version of the function.
-	 */
-	flush_dcache_page(page);
 }

base-commit: 1a1d569a75f3ab2923cb62daf356d102e4df2b86

#3 lore raw
Quoted 28 lines · Samuel Holland

Hi Björn,

Quoted 25 lines · Björn Töpel

From: Björn Töpel <[email protected]>

The XOL (execute out-of-line) buffer is used to single-step the replaced instruction(s) for uprobes. The RISC-V port was missing a proper fence.i (i$ flushing) after constructing the XOL buffer, which can result in incorrect execution of stale/broken instructions.

This was found running the BPF selftests "test_progs:
uprobe_autoattach, attach_probe" on the Spacemit K1/X60, where the uprobes tests randomly blew up.

Fixes 74784081aac8 ("riscv: Add uprobes supported")
---
 arch/riscv/kernel/probes/uprobes.c | 9 +--------
 1 file changed, 1 insertion(+), 8 deletions(-)
arch/riscv/kernel/probes/uprobes.c +0 -0
index 4b3dc8beaf77..4faef92dd771 100644
@@ -176,13 +176,6 @@ void arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr,
 		*(uprobe_opcode_t *)dst = __BUG_INSN_32;
 	}
 
Quoted 6 lines

arch/riscv/kernel/probes/uprobes.c void arch_uprobe_copy_ixol(struct page *page, unsigned long vaddr, · line 179

Quoted 1 line
arch/riscv/kernel/probes/uprobes.c +1 -0
+	flush_icache_range((unsigned long)dst, (unsigned long)dst + len);

This works because flush_icache_range currently ignores the range, but semantically is not quite right, because of the line just above the context that increments dst. If the range was respected, this would only flush the ebreak, not the preceding single-stepped instruction.

Indeed! That was sloppy! I'll spin a v2!

Thank you!
Björn