Skip to contentSkip to conversation

x86/dumpstack: Fix misleading instruction pointer error message

PATCH Superseded Mark Mossberg 2 participants 2 messages Sep 25, 2020

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 x86/dumpstack: Fix misleading instruction pointer error message

Unconditionally printing "Bad RIP value" if copy_code() fails can be misleading for userspace pointers, since copy_code() can fail if the instruction pointer is valid, but the code is paged out. This is because copy_code() calls copy_from_user_nmi() for userspace pointers, which disables page fault handling.

This is reproducible in OOM situations, where it's plausible that the code may be reclaimed in the time between entry into the kernel and when this message is printed. This leaves a misleading log in dmesg that suggests instruction pointer corruption has occurred, which may alarm users.

This patch changes the message printed for userspace pointers to more accurately reflect the possible reasons why the code cannot be dumped.

---
 arch/x86/kernel/dumpstack.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
arch/x86/kernel/dumpstack.c +4 -1
index 48ce44576947..37dbf16c7456 100644
@@ -115,7 +115,10 @@ void show_opcodes(struct pt_regs *regs, const char *loglvl)
 	unsigned long prologue = regs->ip - PROLOGUE_SIZE;
 
 	if (copy_code(regs, opcodes, prologue, sizeof(opcodes))) {
-		printk("%sCode: Bad RIP value.\n", loglvl);
+		if (user_mode(regs))
+			printk("%sCode: Bad RIP value or code paged out.\n", loglvl);
+		else
+			printk("%sCode: Bad RIP value.\n", loglvl);
 	} else {
 		printk("%sCode: %" __stringify(PROLOGUE_SIZE) "ph <%02x> %"
 		       __stringify(EPILOGUE_SIZE) "ph\n", loglvl, opcodes,
Signature · 1 line
2.25.1
maintainer
#1 lore raw
Quoted 25 lines · Mark Mossberg

Unconditionally printing "Bad RIP value" if copy_code() fails can be misleading for userspace pointers, since copy_code() can fail if the instruction pointer is valid, but the code is paged out. This is because copy_code() calls copy_from_user_nmi() for userspace pointers, which disables page fault handling.

This is reproducible in OOM situations, where it's plausible that the code may be reclaimed in the time between entry into the kernel and when this message is printed. This leaves a misleading log in dmesg that suggests instruction pointer corruption has occurred, which may alarm users.

This patch changes the message printed for userspace pointers to more accurately reflect the possible reasons why the code cannot be dumped.

---
 arch/x86/kernel/dumpstack.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
arch/x86/kernel/dumpstack.c +0 -0
index 48ce44576947..37dbf16c7456 100644

arch/x86/kernel/dumpstack.c void show_opcodes(struct pt_regs *regs, const char *loglvl) · line 118

Quoted 4 lines
arch/x86/kernel/dumpstack.c +0 -1
@@ -115,7 +115,10 @@ void show_opcodes(struct pt_regs *regs, const char *loglvl)
 	unsigned long prologue = regs->ip - PROLOGUE_SIZE;
 
 	if (copy_code(regs, opcodes, prologue, sizeof(opcodes))) {
-		printk("%sCode: Bad RIP value.\n", loglvl);

I'd prefer if this thing said exactly what the problem is:

		printk("%sCode: Unable to access opcode bytes at rIP 0x%lx... "

or so.

Signature · 4 lines
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette