OSDN Git Service

bpf: Fix nested bpf_bprintf_prepare with more per-cpu buffers
authorFlorent Revest <revest@chromium.org>
Tue, 11 May 2021 08:10:54 +0000 (10:10 +0200)
committerAlexei Starovoitov <ast@kernel.org>
Tue, 11 May 2021 21:02:33 +0000 (14:02 -0700)
The bpf_seq_printf, bpf_trace_printk and bpf_snprintf helpers share one
per-cpu buffer that they use to store temporary data (arguments to
bprintf). They "get" that buffer with try_get_fmt_tmp_buf and "put" it
by the end of their scope with bpf_bprintf_cleanup.

If one of these helpers gets called within the scope of one of these
helpers, for example: a first bpf program gets called, uses
bpf_trace_printk which calls raw_spin_lock_irqsave which is traced by
another bpf program that calls bpf_snprintf, then the second "get"
fails. Essentially, these helpers are not re-entrant. They would return
-EBUSY and print a warning message once.

This patch triples the number of bprintf buffers to allow three levels
of nesting. This is very similar to what was done for tracepoints in
"9594dc3c7e7 bpf: fix nested bpf tracepoints with per-cpu data"

Fixes: d9c9e4db186a ("bpf: Factorize bpf_trace_printk and bpf_seq_printf")
Reported-by: syzbot+63122d0bc347f18c1884@syzkaller.appspotmail.com
Signed-off-by: Florent Revest <revest@chromium.org>
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/bpf/20210511081054.2125874-1-revest@chromium.org
kernel/bpf/helpers.c

index 5447739..ef658a9 100644 (file)
@@ -696,34 +696,35 @@ static int bpf_trace_copy_string(char *buf, void *unsafe_ptr, char fmt_ptype,
  */
 #define MAX_PRINTF_BUF_LEN     512
 
-struct bpf_printf_buf {
-       char tmp_buf[MAX_PRINTF_BUF_LEN];
+/* Support executing three nested bprintf helper calls on a given CPU */
+struct bpf_bprintf_buffers {
+       char tmp_bufs[3][MAX_PRINTF_BUF_LEN];
 };
-static DEFINE_PER_CPU(struct bpf_printf_buf, bpf_printf_buf);
-static DEFINE_PER_CPU(int, bpf_printf_buf_used);
+static DEFINE_PER_CPU(struct bpf_bprintf_buffers, bpf_bprintf_bufs);
+static DEFINE_PER_CPU(int, bpf_bprintf_nest_level);
 
 static int try_get_fmt_tmp_buf(char **tmp_buf)
 {
-       struct bpf_printf_buf *bufs;
-       int used;
+       struct bpf_bprintf_buffers *bufs;
+       int nest_level;
 
        preempt_disable();
-       used = this_cpu_inc_return(bpf_printf_buf_used);
-       if (WARN_ON_ONCE(used > 1)) {
-               this_cpu_dec(bpf_printf_buf_used);
+       nest_level = this_cpu_inc_return(bpf_bprintf_nest_level);
+       if (WARN_ON_ONCE(nest_level > ARRAY_SIZE(bufs->tmp_bufs))) {
+               this_cpu_dec(bpf_bprintf_nest_level);
                preempt_enable();
                return -EBUSY;
        }
-       bufs = this_cpu_ptr(&bpf_printf_buf);
-       *tmp_buf = bufs->tmp_buf;
+       bufs = this_cpu_ptr(&bpf_bprintf_bufs);
+       *tmp_buf = bufs->tmp_bufs[nest_level - 1];
 
        return 0;
 }
 
 void bpf_bprintf_cleanup(void)
 {
-       if (this_cpu_read(bpf_printf_buf_used)) {
-               this_cpu_dec(bpf_printf_buf_used);
+       if (this_cpu_read(bpf_bprintf_nest_level)) {
+               this_cpu_dec(bpf_bprintf_nest_level);
                preempt_enable();
        }
 }