x86/kprobes: Fix crash when probing CS CALL instructionsgrafted

When using eBPF to probe CS CALL instructions within a function,
a crash can be triggered.

The eBPF tool probes offset 257 of the __hrtimer_run_queues()
function:

<__hrtimer_run_queues+249>:  nopl   0x0(%rax,%rax,1)
<__hrtimer_run_queues+254>:  mov    %r14,%rdi
<__hrtimer_run_queues+257>:  cs call <__x86_indirect_thunk_r12>
<__hrtimer_run_queues+263>:  mov    %eax,%r12d
<__hrtimer_run_queues+266>:  xchg   %ax,%ax
<__hrtimer_run_queues+268>:  mov    %r13,%rdi

Which triggers this crash:

  BUG: unable to handle page fault for address: 00000000000f41c9
  #PF: supervisor write access in kernel mode
  #PF: error_code(0x0002) - not-present page
  PGD 0 P4D 0
  Oops: 0002 [#1] SMP NOPTI
  CPU: 1 PID: 0 Comm: swapper/1 Kdump: loaded Tainted: P
  RIP: 0010:__hrtimer_run_queues+0x106/0x230

Note that __hrtimer_run_queues+0x106 is __hrtimer_run_queues+262, which is
at the 6th byte of the above CS CALL instruction. Since the CS CALL
instruction occupies 6 bytes, the exception occurred in the middle of that
call instruction.

The root cause is that when using eBPF tools to probe in the middle of a
function, a kprobe with INT3 is used as the underlying implementation.

During single-step emulation of the original CALL instruction,
int3_emulate_call() assumes that the probed CALL instruction is 5 bytes
long. However, the actual CS-prefixed CALL instruction occupies 6 bytes,
so it constructs an incorrect exception return address. When the CPU
returns from the kprobe handler, the next instruction to be executed is at
the address of the last byte of that CS CALL instruction. Coincidentally,
starting from that address, the CPU fetches and decodes a completely
different instruction, which ultimately triggers a kernel crash.

Fix the issue by using the actual instruction length obtained from
the instruction decoder when constructing the exception return
address, rather than relying on the hardcoded CALL_INSN_SIZE macro.

[ mingo: Refined the changelog ]

Fixes: 6256e668b7af ("x86/kprobes: Use int3 instead of debug trap for single-step")
Suggested-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Jinke Han <jinkehan@didiglobal.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Acked-by: Yafang Shao <laoar.shao@gmail.com>
Acked-by: Borislav Petkov <bp@alien8.de>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: https://patch.msgid.link/20260908073742.GA10517@didi-ThinkCentre-M920t-N000

1 files changed, 188 insertions, 0 deletions

diff --git a/arch/sh/mm/cache-sh2a.c b/arch/sh/mm/cache-sh2a.c
new file mode 100644
index 000000000..4f7739e70
--- /dev/null
+++ b/arch/sh/mm/cache-sh2a.c
@@ -0,0 +1,188 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * arch/sh/mm/cache-sh2a.c
+ *
+ * Copyright (C) 2008 Yoshinori Sato
+ */
+
+#include <linux/init.h>
+#include <linux/mm.h>
+
+#include <asm/cache.h>
+#include <asm/addrspace.h>
+#include <asm/processor.h>
+#include <asm/cacheflush.h>
+#include <asm/io.h>
+
+/*
+ * The maximum number of pages we support up to when doing ranged dcache
+ * flushing. Anything exceeding this will simply flush the dcache in its
+ * entirety.
+ */
+#define MAX_OCACHE_PAGES	32
+#define MAX_ICACHE_PAGES	32
+
+#ifdef CONFIG_CACHE_WRITEBACK
+static void sh2a_flush_oc_line(unsigned long v, int way)
+{
+	unsigned long addr = (v & 0x000007f0) | (way << 11);
+	unsigned long data;
+
+	data = __raw_readl(CACHE_OC_ADDRESS_ARRAY | addr);
+	if ((data & CACHE_PHYSADDR_MASK) == (v & CACHE_PHYSADDR_MASK)) {
+		data &= ~SH_CACHE_UPDATED;
+		__raw_writel(data, CACHE_OC_ADDRESS_ARRAY | addr);
+	}
+}
+#endif
+
+static void sh2a_invalidate_line(unsigned long cache_addr, unsigned long v)
+{
+	/* Set associative bit to hit all ways */
+	unsigned long addr = (v & 0x000007f0) | SH_CACHE_ASSOC;
+	__raw_writel((addr & CACHE_PHYSADDR_MASK), cache_addr | addr);
+}
+
+/*
+ * Write back the dirty D-caches, but not invalidate them.
+ */
+static void sh2a__flush_wback_region(void *start, int size)
+{
+#ifdef CONFIG_CACHE_WRITEBACK
+	unsigned long v;
+	unsigned long begin, end;
+	unsigned long flags;
+	int nr_ways;
+
+	begin = (unsigned long)start & ~(L1_CACHE_BYTES-1);
+	end = ((unsigned long)start + size + L1_CACHE_BYTES-1)
+		& ~(L1_CACHE_BYTES-1);
+	nr_ways = current_cpu_data.dcache.ways;
+
+	local_irq_save(flags);
+	jump_to_uncached();
+
+	/* If there are too many pages then flush the entire cache */
+	if (((end - begin) >> PAGE_SHIFT) >= MAX_OCACHE_PAGES) {
+		begin = CACHE_OC_ADDRESS_ARRAY;
+		end = begin + (nr_ways * current_cpu_data.dcache.way_size);
+
+		for (v = begin; v < end; v += L1_CACHE_BYTES) {
+			unsigned long data = __raw_readl(v);
+			if (data & SH_CACHE_UPDATED)
+				__raw_writel(data & ~SH_CACHE_UPDATED, v);
+		}
+	} else {
+		int way;
+		for (way = 0; way < nr_ways; way++) {
+			for (v = begin; v < end; v += L1_CACHE_BYTES)
+				sh2a_flush_oc_line(v, way);
+		}
+	}
+
+	back_to_cached();
+	local_irq_restore(flags);
+#endif
+}
+
+/*
+ * Write back the dirty D-caches and invalidate them.
+ */
+static void sh2a__flush_purge_region(void *start, int size)
+{
+	unsigned long v;
+	unsigned long begin, end;
+	unsigned long flags;
+
+	begin = (unsigned long)start & ~(L1_CACHE_BYTES-1);
+	end = ((unsigned long)start + size + L1_CACHE_BYTES-1)
+		& ~(L1_CACHE_BYTES-1);
+
+	local_irq_save(flags);
+	jump_to_uncached();
+
+	for (v = begin; v < end; v+=L1_CACHE_BYTES) {
+#ifdef CONFIG_CACHE_WRITEBACK
+		int way;
+		int nr_ways = current_cpu_data.dcache.ways;
+		for (way = 0; way < nr_ways; way++)
+			sh2a_flush_oc_line(v, way);
+#endif
+		sh2a_invalidate_line(CACHE_OC_ADDRESS_ARRAY, v);
+	}
+
+	back_to_cached();
+	local_irq_restore(flags);
+}
+
+/*
+ * Invalidate the D-caches, but no write back please
+ */
+static void sh2a__flush_invalidate_region(void *start, int size)
+{
+	unsigned long v;
+	unsigned long begin, end;
+	unsigned long flags;
+
+	begin = (unsigned long)start & ~(L1_CACHE_BYTES-1);
+	end = ((unsigned long)start + size + L1_CACHE_BYTES-1)
+		& ~(L1_CACHE_BYTES-1);
+
+	local_irq_save(flags);
+	jump_to_uncached();
+
+	/* If there are too many pages then just blow the cache */
+	if (((end - begin) >> PAGE_SHIFT) >= MAX_OCACHE_PAGES) {
+		__raw_writel(__raw_readl(SH_CCR) | CCR_OCACHE_INVALIDATE,
+			     SH_CCR);
+	} else {
+		for (v = begin; v < end; v += L1_CACHE_BYTES)
+			sh2a_invalidate_line(CACHE_OC_ADDRESS_ARRAY, v);
+	}
+
+	back_to_cached();
+	local_irq_restore(flags);
+}
+
+/*
+ * Write back the range of D-cache, and purge the I-cache.
+ */
+static void sh2a_flush_icache_range(void *args)
+{
+	struct flusher_data *data = args;
+	unsigned long start, end;
+	unsigned long v;
+	unsigned long flags;
+
+	start = data->addr1 & ~(L1_CACHE_BYTES-1);
+	end = (data->addr2 + L1_CACHE_BYTES-1) & ~(L1_CACHE_BYTES-1);
+
+#ifdef CONFIG_CACHE_WRITEBACK
+	sh2a__flush_wback_region((void *)start, end-start);
+#endif
+
+	local_irq_save(flags);
+	jump_to_uncached();
+
+	/* I-Cache invalidate */
+	/* If there are too many pages then just blow the cache */
+	if (((end - start) >> PAGE_SHIFT) >= MAX_ICACHE_PAGES) {
+		__raw_writel(__raw_readl(SH_CCR) | CCR_ICACHE_INVALIDATE,
+			     SH_CCR);
+	} else {
+		for (v = start; v < end; v += L1_CACHE_BYTES)
+			sh2a_invalidate_line(CACHE_IC_ADDRESS_ARRAY, v);
+	}
+
+	back_to_cached();
+	local_irq_restore(flags);
+}
+
+void __init sh2a_cache_init(void)
+{
+	local_flush_icache_range	= sh2a_flush_icache_range;
+
+	__flush_wback_region		= sh2a__flush_wback_region;
+	__flush_purge_region		= sh2a__flush_purge_region;
+	__flush_invalidate_region	= sh2a__flush_invalidate_region;
+}