On Mon, Jul 24, 2023 at 10:16:56AM +0800, Zheng Yejian wrote: > commit d5a821896360cc8b93a15bd888fabc858c038dc0 upstream. > > kmemleak reports: > unreferenced object 0xffff88814d14e200 (size 256): > comm "cat", pid 336, jiffies 4294871818 (age 779.490s) > hex dump (first 32 bytes): > 04 00 01 03 00 00 00 00 08 00 00 00 00 00 00 00 ................ > 0c d8 c8 9b ff ff ff ff 04 5a ca 9b ff ff ff ff .........Z...... > backtrace: > [<ffffffff9bdff18f>] __kmalloc+0x4f/0x140 > [<ffffffff9bc9238b>] trace_find_next_entry+0xbb/0x1d0 > [<ffffffff9bc9caef>] trace_print_lat_context+0xaf/0x4e0 > [<ffffffff9bc94490>] print_trace_line+0x3e0/0x950 > [<ffffffff9bc95499>] tracing_read_pipe+0x2d9/0x5a0 > [<ffffffff9bf03a43>] vfs_read+0x143/0x520 > [<ffffffff9bf04c2d>] ksys_read+0xbd/0x160 > [<ffffffff9d0f0edf>] do_syscall_64+0x3f/0x90 > [<ffffffff9d2000aa>] entry_SYSCALL_64_after_hwframe+0x6e/0xd8 > > when reading file 'trace_pipe', 'iter->temp' is allocated or relocated > in trace_find_next_entry() but not freed before 'trace_pipe' is closed. > > To fix it, free 'iter->temp' in tracing_release_pipe(). > > Link: https://lore.kernel.org/linux-trace-kernel/20230713141435.1133021-1-zhengyejian1@xxxxxxxxxx > > Cc: stable@xxxxxxxxxxxxxxx > Fixes: ff895103a84ab ("tracing: Save off entry when peeking at next entry") > Signed-off-by: Zheng Yejian <zhengyejian1@xxxxxxxxxx> > Signed-off-by: Steven Rostedt (Google) <rostedt@xxxxxxxxxxx> > [Fix conflict due to lack of 649e72070cbbb8600eb823833e4748f5a0815116] > Signed-off-by: Zheng Yejian <zhengyejian1@xxxxxxxxxx> > --- > kernel/trace/trace.c | 1 + > 1 file changed, 1 insertion(+) > > diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c > index 70526400e05c..8d005c7ccfae 100644 > --- a/kernel/trace/trace.c > +++ b/kernel/trace/trace.c > @@ -6250,6 +6250,7 @@ static int tracing_release_pipe(struct inode *inode, struct file *file) > mutex_unlock(&trace_types_lock); > > free_cpumask_var(iter->started); > + kfree(iter->temp); > mutex_destroy(&iter->mutex); > kfree(iter); > > -- > 2.25.1 > Now queued up, thanks. greg k-h
