Skip to contentSkip to conversation

tracing: A few more fixes for 5.7

PATCH In review Steven Rostedt 2 participants 5 messages May 14, 2020 – May 18, 2020

Subsystems (6)
Stable branch x86 Tracing x86 MM +2 more subsystems
lore
Share & tools
lore

Steven Rostedt

author maintainer
#0 lore raw

cover tracing: A few more fixes for 5.7

Steven Rostedt (VMware) (3):
x86/ftrace: Have ftrace trampolines turn read-only at the end of system boot up
ring-buffer: Don't deactivate the ring buffer on failed iterator reads
ring-buffer: Remove all BUG() calls

----

 arch/x86/include/asm/ftrace.h  |  6 ++++++
 arch/x86/kernel/ftrace.c       | 29 ++++++++++++++++++++++++++++-
 arch/x86/mm/init_64.c          |  3 +++
 include/linux/ftrace.h         | 23 +++++++++++++++++++++++
 kernel/trace/ftrace_internal.h | 22 ----------------------
 kernel/trace/ring_buffer.c     | 34 +++++++++++++---------------------
 6 files changed, 73 insertions(+), 44 deletions(-)
author maintainer
#1 lore raw

PATCH 1/3 x86/ftrace: Have ftrace trampolines turn read-only at the end of system boot up

From: "Steven Rostedt (VMware)" <[email protected]>

Booting one of my machines, it triggered the following crash:

Kernel/User page tables isolation: enabled
ftrace: allocating 36577 entries in 143 pages
Starting tracer 'function'

 BUG: unable to handle page fault for address: ffffffffa000005c
 #PF: supervisor write access in kernel mode
 #PF: error_code(0x0003) - permissions violation
 PGD 2014067 P4D 2014067 PUD 2015063 PMD 7b253067 PTE 7b252061
 Oops: 0003 [#1] PREEMPT SMP PTI
 CPU: 0 PID: 0 Comm: swapper Not tainted 5.4.0-test+ #24
 Hardware name: To Be Filled By O.E.M. To Be Filled By O.E.M./To be filled by O.E.M., BIOS SDBLI944.86P 05/08/2007
 RIP: 0010:text_poke_early+0x4a/0x58
 Code: 34 24 48 89 54 24 08 e8 bf 72 0b 00 48 8b 34 24 48 8b 4c 24 08 84 c0 74 0b 48 89 df f3 a4 48 83 c4 10 5b c3 9c 58 fa 48 89 df <f3> a4 50 9d 48 83 c4 10 5b e9 d6 f9 ff ff

0 41 57 49
RSP: 0000:ffffffff82003d38 EFLAGS: 00010046
RAX: 0000000000000046 RBX: ffffffffa000005c RCX: 0000000000000005
RDX: 0000000000000005 RSI: ffffffff825b9a90 RDI: ffffffffa000005c
RBP: ffffffffa000005c R08: 0000000000000000 R09: ffffffff8206e6e0
R10: ffff88807b01f4c0 R11: ffffffff8176c106 R12: ffffffff8206e6e0
R13: ffffffff824f2440 R14: 0000000000000000 R15: ffffffff8206eac0
FS: 0000000000000000(0000) GS:ffff88807d400000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: ffffffffa000005c CR3: 0000000002012000 CR4: 00000000000006b0

 Call Trace:
  text_poke_bp+0x27/0x64
  ? mutex_lock+0x36/0x5d
  arch_ftrace_update_trampoline+0x287/0x2d5
  ? ftrace_replace_code+0x14b/0x160
  ? ftrace_update_ftrace_func+0x65/0x6c
  __register_ftrace_function+0x6d/0x81
  ftrace_startup+0x23/0xc1
  register_ftrace_function+0x20/0x37
  func_set_flag+0x59/0x77
  __set_tracer_option.isra.19+0x20/0x3e
  trace_set_options+0xd6/0x13e
  apply_trace_boot_options+0x44/0x6d
  register_tracer+0x19e/0x1ac
  early_trace_init+0x21b/0x2c9
  start_kernel+0x241/0x518
  ? load_ucode_intel_bsp+0x21/0x52
  secondary_startup_64+0xa4/0xb0

I was able to trigger it on other machines, when I added to the kernel command line of both "ftrace=function" and "trace_options=func_stack_trace".

The cause is the "ftrace=function" would register the function tracer and create a trampoline, and it will set it as executable and read-only. Then the "trace_options=func_stack_trace" would then update the same trampoline to include the stack tracer version of the function tracer. But since the trampoline already exists, it updates it with text_poke_bp(). The problem is that text_poke_bp() called while system_state == SYSTEM_BOOTING, it will simply do a memcpy() and not the page mapping, as it would think that the text is still read-write. But in this case it is not, and we take a fault and crash.

Instead, lets keep the ftrace trampolines read-write during boot up, and then when the kernel executable text is set to read-only, the ftrace trampolines get set to read-only as well.

Cc Ingo Molnar <[email protected]>
Cc Thomas Gleixner <[email protected]>
Cc Andy Lutomirski <[email protected]>
Cc Borislav Petkov <[email protected]>
Cc Josh Poimboeuf <[email protected]>
Cc "H. Peter Anvin" <[email protected]>
Fixes 768ae4406a5c ("x86/ftrace: Use text_poke()")
---
 arch/x86/include/asm/ftrace.h  |  6 ++++++
 arch/x86/kernel/ftrace.c       | 29 ++++++++++++++++++++++++++++-
 arch/x86/mm/init_64.c          |  3 +++
 include/linux/ftrace.h         | 23 +++++++++++++++++++++++
 kernel/trace/ftrace_internal.h | 22 ----------------------
 5 files changed, 60 insertions(+), 23 deletions(-)
arch/x86/include/asm/ftrace.h +6 -0
index 85be2f506272..89af0d2c62aa 100644
@@ -56,6 +56,12 @@ struct dyn_arch_ftrace {
 
 #ifndef __ASSEMBLY__
 
+#if defined(CONFIG_FUNCTION_TRACER) && defined(CONFIG_DYNAMIC_FTRACE)
+extern void set_ftrace_ops_ro(void);
+#else
+static inline void set_ftrace_ops_ro(void) { }
+#endif
+
 #define ARCH_HAS_SYSCALL_MATCH_SYM_NAME
 static inline bool arch_syscall_match_sym_name(const char *sym, const char *name)
 {
arch/x86/kernel/ftrace.c +28 -1
index 37a0aeaf89e7..b0e641793be4 100644
@@ -407,7 +407,8 @@ create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size)
 
 	set_vm_flush_reset_perms(trampoline);
 
-	set_memory_ro((unsigned long)trampoline, npages);
+	if (likely(system_state != SYSTEM_BOOTING))
+		set_memory_ro((unsigned long)trampoline, npages);
 	set_memory_x((unsigned long)trampoline, npages);
 	return (unsigned long)trampoline;
 fail:
@@ -415,6 +416,32 @@ create_trampoline(struct ftrace_ops *ops, unsigned int *tramp_size)
 	return 0;
 }
 
+void set_ftrace_ops_ro(void)
+{
+	struct ftrace_ops *ops;
+	unsigned long start_offset;
+	unsigned long end_offset;
+	unsigned long npages;
+	unsigned long size;
+
+	do_for_each_ftrace_op(ops, ftrace_ops_list) {
+		if (!(ops->flags & FTRACE_OPS_FL_ALLOC_TRAMP))
+			continue;
+
+		if (ops->flags & FTRACE_OPS_FL_SAVE_REGS) {
+			start_offset = (unsigned long)ftrace_regs_caller;
+			end_offset = (unsigned long)ftrace_regs_caller_end;
+		} else {
+			start_offset = (unsigned long)ftrace_caller;
+			end_offset = (unsigned long)ftrace_epilogue;
+		}
+		size = end_offset - start_offset;
+		size = size + RET_SIZE + sizeof(void *);
+		npages = DIV_ROUND_UP(size, PAGE_SIZE);
+		set_memory_ro((unsigned long)ops->trampoline, npages);
+	} while_for_each_ftrace_op(ops);
+}
+
 static unsigned long calc_trampoline_call_offset(bool save_regs)
 {
 	unsigned long start_offset;
arch/x86/mm/init_64.c +3 -0
index 3b289c2f75cd..8b5f73f5e207 100644
@@ -54,6 +54,7 @@
 #include <asm/init.h>
 #include <asm/uv/uv.h>
 #include <asm/setup.h>
+#include <asm/ftrace.h>
 
 #include "mm_internal.h"
 
@@ -1291,6 +1292,8 @@ void mark_rodata_ro(void)
 	all_end = roundup((unsigned long)_brk_end, PMD_SIZE);
 	set_memory_nx(text_end, (all_end - text_end) >> PAGE_SHIFT);
 
+	set_ftrace_ops_ro();
+
 #ifdef CONFIG_CPA_DEBUG
 	printk(KERN_INFO "Testing CPA: undo %lx-%lx\n", start, end);
 	set_memory_rw(start, (end-start) >> PAGE_SHIFT);
include/linux/ftrace.h +23 -0
index db95244a62d4..ab4bd15cbcdb 100644
@@ -210,6 +210,29 @@ struct ftrace_ops {
 #endif
 };
 
+extern struct ftrace_ops __rcu *ftrace_ops_list;
+extern struct ftrace_ops ftrace_list_end;
+
+/*
+ * Traverse the ftrace_global_list, invoking all entries.  The reason that we
+ * can use rcu_dereference_raw_check() is that elements removed from this list
+ * are simply leaked, so there is no need to interact with a grace-period
+ * mechanism.  The rcu_dereference_raw_check() calls are needed to handle
+ * concurrent insertions into the ftrace_global_list.
+ *
+ * Silly Alpha and silly pointer-speculation compiler optimizations!
+ */
+#define do_for_each_ftrace_op(op, list)			\
+	op = rcu_dereference_raw_check(list);			\
+	do
+
+/*
+ * Optimized for just a single item in the list (as that is the normal case).
+ */
+#define while_for_each_ftrace_op(op)				\
+	while (likely(op = rcu_dereference_raw_check((op)->next)) &&	\
+	       unlikely((op) != &ftrace_list_end))
+
 /*
  * Type of the current tracing.
  */
kernel/trace/ftrace_internal.h +0 -22
index 0456e0a3dab1..382775edf690 100644
@@ -4,28 +4,6 @@
 
 #ifdef CONFIG_FUNCTION_TRACER
 
-/*
- * Traverse the ftrace_global_list, invoking all entries.  The reason that we
- * can use rcu_dereference_raw_check() is that elements removed from this list
- * are simply leaked, so there is no need to interact with a grace-period
- * mechanism.  The rcu_dereference_raw_check() calls are needed to handle
- * concurrent insertions into the ftrace_global_list.
- *
- * Silly Alpha and silly pointer-speculation compiler optimizations!
- */
-#define do_for_each_ftrace_op(op, list)			\
-	op = rcu_dereference_raw_check(list);			\
-	do
-
-/*
- * Optimized for just a single item in the list (as that is the normal case).
- */
-#define while_for_each_ftrace_op(op)				\
-	while (likely(op = rcu_dereference_raw_check((op)->next)) &&	\
-	       unlikely((op) != &ftrace_list_end))
-
-extern struct ftrace_ops __rcu *ftrace_ops_list;
-extern struct ftrace_ops ftrace_list_end;
 extern struct mutex ftrace_lock;
 extern struct ftrace_ops global_ops;
 
Signature · 1 line
2.26.2
maintainer
#2 lore raw

Re: [for-linus][PATCH 1/3] x86/ftrace: Have ftrace trampolines turn read-only at the end of system boot up

Review

Quoted 2 lines · Steven Rostedt
(quoted diff) +2 -0
+			start_offset = (unsigned long)ftrace_caller;
+			end_offset = (unsigned long)ftrace_epilogue;

---
Subject: x86/ftrace: Fix compile error

When building x86-64 kernels, my compiler is sad about a missing symbol.

Fixes 59566b0b622e ("x86/ftrace: Have ftrace trampolines turn read-only at the end of system boot up")
---
 arch/x86/kernel/ftrace.c | 1 +
 1 file changed, 1 insertion(+)
arch/x86/kernel/ftrace.c +1 -0
index f8917a6f25b7..1cf7d69402e2 100644
@@ -286,6 +286,7 @@ extern void ftrace_regs_caller_ret(void);
 extern void ftrace_caller_end(void);
 extern void ftrace_caller_op_ptr(void);
 extern void ftrace_regs_caller_op_ptr(void);
+extern void ftrace_epilogue(void);
 
 /* movq function_trace_op(%rip), %rdx */
 /* 0x48 0x8b 0x15 <offset-to-ftrace_trace_op (4 bytes)> */
author maintainer
#3 lore raw

PATCH 2/3 ring-buffer: Dont deactivate the ring buffer on failed iterator reads

In reply to Steven Rostedt

From: "Steven Rostedt (VMware)" <[email protected]>

If the function tracer is running and the trace file is read (which uses the ring buffer iterator), the iterator can get in sync with the writes, and caues it to fail to find a page with content it can read three times. This causes a warning and deactivation of the ring buffer code.

Looking at the other cases of failure to get an event, it appears that there's a chance that the writer could cause them too. Since the iterator is a "best effort" to read the ring buffer if there's an active writer (the consumer reader is made for this case "see trace_pipe"), if it fails to get an event after three tries, simply give up and return NULL. Don't warn, nor disable the ring buffer on this failure.

Fixes ff84c50cfb4b ("ring-buffer: Do not die if rb_iter_peek() fails more than thrice")
---
 kernel/trace/ring_buffer.c | 22 +++++++---------------
 1 file changed, 7 insertions(+), 15 deletions(-)
kernel/trace/ring_buffer.c +7 -15
index 6f0b42ceeb00..448d5f528764 100644
@@ -4034,7 +4034,6 @@ rb_iter_peek(struct ring_buffer_iter *iter, u64 *ts)
 	struct ring_buffer_per_cpu *cpu_buffer;
 	struct ring_buffer_event *event;
 	int nr_loops = 0;
-	bool failed = false;
 
 	if (ts)
 		*ts = 0;
@@ -4056,19 +4055,14 @@ rb_iter_peek(struct ring_buffer_iter *iter, u64 *ts)
 		return NULL;
 
 	/*
-	 * We repeat when a time extend is encountered or we hit
-	 * the end of the page. Since the time extend is always attached
-	 * to a data event, we should never loop more than three times.
-	 * Once for going to next page, once on time extend, and
-	 * finally once to get the event.
-	 * We should never hit the following condition more than thrice,
-	 * unless the buffer is very small, and there's a writer
-	 * that is causing the reader to fail getting an event.
+	 * As the writer can mess with what the iterator is trying
+	 * to read, just give up if we fail to get an event after
+	 * three tries. The iterator is not as reliable when reading
+	 * the ring buffer with an active write as the consumer is.
+	 * Do not warn if the three failures is reached.
 	 */
-	if (++nr_loops > 3) {
-		RB_WARN_ON(cpu_buffer, !failed);
+	if (++nr_loops > 3)
 		return NULL;
-	}
 
 	if (rb_per_cpu_empty(cpu_buffer))
 		return NULL;
@@ -4079,10 +4073,8 @@ rb_iter_peek(struct ring_buffer_iter *iter, u64 *ts)
 	}
 
 	event = rb_iter_head_event(iter);
-	if (!event) {
-		failed = true;
+	if (!event)
 		goto again;
-	}
 
 	switch (event->type_len) {
 	case RINGBUF_TYPE_PADDING:
Signature · 1 line
2.26.2
author maintainer
#4 lore raw

PATCH 3/3 ring-buffer: Remove all BUG() calls

In reply to Steven Rostedt

From: "Steven Rostedt (VMware)" <[email protected]>

There's a lot of checks to make sure the ring buffer is working, and if an anomaly is detected, it safely shuts itself down. But there's a few cases that it will call BUG(), which defeats the point of being safe (it crashes the kernel when an anomaly is found!). There's no reason for them. Switch them all to either WARN_ON_ONCE() (when no ring buffer descriptor is present), or to RB_WARN_ON() (when a ring buffer descriptor is present).

---
 kernel/trace/ring_buffer.c | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)
kernel/trace/ring_buffer.c +6 -6
index 448d5f528764..b8e1ca48be50 100644
@@ -193,7 +193,7 @@ rb_event_length(struct ring_buffer_event *event)
 	case RINGBUF_TYPE_DATA:
 		return rb_event_data_length(event);
 	default:
-		BUG();
+		WARN_ON_ONCE(1);
 	}
 	/* not hit */
 	return 0;
@@ -249,7 +249,7 @@ rb_event_data(struct ring_buffer_event *event)
 {
 	if (extended_time(event))
 		event = skip_time_extend(event);
-	BUG_ON(event->type_len > RINGBUF_TYPE_DATA_TYPE_LEN_MAX);
+	WARN_ON_ONCE(event->type_len > RINGBUF_TYPE_DATA_TYPE_LEN_MAX);
 	/* If length is in len field, then array[0] has the data */
 	if (event->type_len)
 		return (void *)&event->array[0];
@@ -3727,7 +3727,7 @@ rb_update_read_stamp(struct ring_buffer_per_cpu *cpu_buffer,
 		return;
 
 	default:
-		BUG();
+		RB_WARN_ON(cpu_buffer, 1);
 	}
 	return;
 }
@@ -3757,7 +3757,7 @@ rb_update_iter_read_stamp(struct ring_buffer_iter *iter,
 		return;
 
 	default:
-		BUG();
+		RB_WARN_ON(iter->cpu_buffer, 1);
 	}
 	return;
 }
@@ -4020,7 +4020,7 @@ rb_buffer_peek(struct ring_buffer_per_cpu *cpu_buffer, u64 *ts,
 		return event;
 
 	default:
-		BUG();
+		RB_WARN_ON(cpu_buffer, 1);
 	}
 
 	return NULL;
@@ -4109,7 +4109,7 @@ rb_iter_peek(struct ring_buffer_iter *iter, u64 *ts)
 		return event;
 
 	default:
-		BUG();
+		RB_WARN_ON(cpu_buffer, 1);
 	}
 
 	return NULL;
Signature · 1 line
2.26.2