Skip to contentSkip to conversation

Fix Compilation Warning in kernel/trace/trace.c

PATCH NAK Devendra Naga 2 participants 3 messages Dec 16, 2011 1 message not in the archive Tracing

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Fix Compilation Warning in kernel/trace/trace.c

there was a compilation warning saying
kernel/trace/trace.c:3644:8: warning: ‘page2’ may be used uninitialized in this function

The page2 pointer is used only when nr_pages = 2.

Assigned a null value to the page2 to get rid of warning.

---
 kernel/trace/trace.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
kernel/trace/trace.c +1 -1
index f2bd275..80dfd92 100644
@@ -3641,7 +3641,7 @@ tracing_mark_write(struct file *filp, const char __user *ubuf,
 	int nr_pages = 1;
 	ssize_t written;
 	void *page1;
-	void *page2;
+	void *page2 = NULL;
 	int offset;
 	int size;
 	int len;
Signature · 1 line
1.7.4.1
maintainer Nacked-by on v1 1/1 NAK
#1 lore raw
Quoted 2 lines · Devendra Naga

there was a compilation warning saying
kernel/trace/trace.c:3644:8: warning: ‘page2’ may be used uninitialized in this function

And I also nacked this same fix here:

https://lkml.org/lkml/2011/11/28/287

Quoted 4 lines

The page2 pointer is used only when nr_pages = 2.

Assigned a null value to the page2 to get rid of warning.

Please please please, do not just fix warnings without actually looking at the code.

1) this is a bug in the compiler, which happens to be fixed in newer compilers (which does not produce this warning). It just happens that gcc couldn't tell the simple usage of:

	if (nr_pages == 2)
		page2 = kmap_atomic(pages[1]);

and later:

	if (nr_pages == 2) {
		len = PAGE_SIZE - offset;
		memcpy(&entry->buf, page1 + offset, len);
		memcpy(&entry->buf[len], page2, cnt - len);

and finally:

	if (nr_pages == 2)
		kunmap_atomic(page2);

nr_pages never changes during this code. Gcc probably got confused because on error exit, it does change (to unmap everything).

2) But your fix is still wrong!

Initializing it to NULL is wrong! Because it would hide a kernel bug if page2 was really used uninitialized. As passing NULL as the source of memcpy will panic the kernel.

Again, do not just randomly post fixes to gcc warnings without understanding the real fix.

Thanks,

-- Steve

Quoted 19 lines
---
 kernel/trace/trace.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)
kernel/trace/trace.c +1 -1
index f2bd275..80dfd92 100644
@@ -3641,7 +3641,7 @@ tracing_mark_write(struct file *filp, const char __user *ubuf,
 	int nr_pages = 1;
 	ssize_t written;
 	void *page1;
-	void *page2;
+	void *page2 = NULL;
 	int offset;
 	int size;
 	int len;