cryptospore / rpms / qemu-kvm

Forked from rpms/qemu-kvm 2 years ago
Clone
ddf19c
From 8f6311159977b8ee4b78172caa411d3cee4d2ae5 Mon Sep 17 00:00:00 2001
ddf19c
From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
ddf19c
Date: Tue, 14 Jan 2020 20:23:30 +0000
ddf19c
Subject: [PATCH 4/5] usbredir: Prevent recursion in usbredir_write
ddf19c
MIME-Version: 1.0
ddf19c
Content-Type: text/plain; charset=UTF-8
ddf19c
Content-Transfer-Encoding: 8bit
ddf19c
ddf19c
RH-Author: Dr. David Alan Gilbert <dgilbert@redhat.com>
ddf19c
Message-id: <20200114202331.51831-2-dgilbert@redhat.com>
ddf19c
Patchwork-id: 93344
ddf19c
O-Subject: [RHEL-AV-8.2.0 qemu-kvm PATCH 1/2] usbredir: Prevent recursion in usbredir_write
ddf19c
Bugzilla: 1790844
ddf19c
RH-Acked-by: Peter Xu <peterx@redhat.com>
ddf19c
RH-Acked-by: Philippe Mathieu-Daudé <philmd@redhat.com>
ddf19c
RH-Acked-by: Gerd Hoffmann <kraxel@redhat.com>
ddf19c
ddf19c
From: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
ddf19c
ddf19c
I've got a case where usbredir_write manages to call back into itself
ddf19c
via spice; this patch causes the recursion to fail (0 bytes) the write;
ddf19c
this seems to avoid the deadlock I was previously seeing.
ddf19c
ddf19c
I can't say I fully understand the interaction of usbredir and spice;
ddf19c
but there are a few similar guards in spice and usbredir
ddf19c
to catch other cases especially onces also related to spice_server_char_device_wakeup
ddf19c
ddf19c
This case seems to be triggered by repeated migration+repeated
ddf19c
reconnection of the viewer; but my debugging suggests the migration
ddf19c
finished before this hits.
ddf19c
ddf19c
The backtrace of the hang looks like:
ddf19c
  reds_handle_ticket
ddf19c
  reds_handle_other_links
ddf19c
  reds_channel_do_link
ddf19c
  red_channel_connect
ddf19c
  spicevmc_connect
ddf19c
  usbredir_create_parser
ddf19c
  usbredirparser_do_write
ddf19c
  usbredir_write
ddf19c
  qemu_chr_fe_write
ddf19c
  qemu_chr_write
ddf19c
  qemu_chr_write_buffer
ddf19c
  spice_chr_write
ddf19c
  spice_server_char_device_wakeup
ddf19c
  red_char_device_wakeup
ddf19c
  red_char_device_write_to_device
ddf19c
  vmc_write
ddf19c
  usbredirparser_do_write
ddf19c
  usbredir_write
ddf19c
  qemu_chr_fe_write
ddf19c
  qemu_chr_write
ddf19c
  qemu_chr_write_buffer
ddf19c
  qemu_mutex_lock_impl
ddf19c
ddf19c
and we fail as we land through qemu_chr_write_buffer's lock
ddf19c
twice.
ddf19c
ddf19c
Bug: https://bugzilla.redhat.com/show_bug.cgi?id=1752320
ddf19c
ddf19c
Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
ddf19c
Message-Id: <20191218113012.13331-1-dgilbert@redhat.com>
ddf19c
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
ddf19c
(cherry picked from commit 394642a8d3742c885e397d5bb5ee0ec40743cdc6)
ddf19c
Signed-off-by: Danilo C. L. de Paula <ddepaula@redhat.com>
ddf19c
---
ddf19c
 hw/usb/redirect.c | 9 +++++++++
ddf19c
 1 file changed, 9 insertions(+)
ddf19c
ddf19c
diff --git a/hw/usb/redirect.c b/hw/usb/redirect.c
ddf19c
index e0f5ca6..97f2c3a 100644
ddf19c
--- a/hw/usb/redirect.c
ddf19c
+++ b/hw/usb/redirect.c
ddf19c
@@ -113,6 +113,7 @@ struct USBRedirDevice {
ddf19c
     /* Properties */
ddf19c
     CharBackend cs;
ddf19c
     bool enable_streams;
ddf19c
+    bool in_write;
ddf19c
     uint8_t debug;
ddf19c
     int32_t bootindex;
ddf19c
     char *filter_str;
ddf19c
@@ -290,6 +291,13 @@ static int usbredir_write(void *priv, uint8_t *data, int count)
ddf19c
         return 0;
ddf19c
     }
ddf19c
 
ddf19c
+    /* Recursion check */
ddf19c
+    if (dev->in_write) {
ddf19c
+        DPRINTF("usbredir_write recursion\n");
ddf19c
+        return 0;
ddf19c
+    }
ddf19c
+    dev->in_write = true;
ddf19c
+
ddf19c
     r = qemu_chr_fe_write(&dev->cs, data, count);
ddf19c
     if (r < count) {
ddf19c
         if (!dev->watch) {
ddf19c
@@ -300,6 +308,7 @@ static int usbredir_write(void *priv, uint8_t *data, int count)
ddf19c
             r = 0;
ddf19c
         }
ddf19c
     }
ddf19c
+    dev->in_write = false;
ddf19c
     return r;
ddf19c
 }
ddf19c
 
ddf19c
-- 
ddf19c
1.8.3.1
ddf19c