sergiotarxz/nuttx
sergiotarxz/nuttx
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Fix for recvfrom() hang when the new CONFIG_NET_TCP_RECVDELAY is set to zero (from Max Holtzberg)

Browse Source 
git-svn-id: svn://svn.code.sf.net/p/nuttx/code/trunk@5148 42af7a65-404d-4744-a932-0658087f49c3
This commit is contained in:
patacongo 2012-09-13 20:37:24 +00:00
parent f6b727fa27
commit 3e7c5c1b43
2 changed files with 22 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
ChangeLog
View File

@ -3337,4 +3337,8 @@
* include/nuttx/usb/usb.h, arch/*/src/*usb.c, and arch/*/src/*otg*.c:
Add hooks to to use common, external DMA buffer allocation
implementation.
* net/recvfrom.c: Don't block in recvfrom if (1) read-ahead buffering
is enabled and (2) some data was obtained from read-ahead buffers.
Blocking is a bad idea in that case because there is no timeout!
(submitted by Max Holtzberg).

18
net/recvfrom.c
View File

@ -1079,7 +1079,25 @@ static ssize_t tcp_recvfrom(FAR struct socket *psock, FAR void *buf, size_t len,
else
#endif
/* We get here when we we decide that we need to setup the wait for incoming
* TCP/IP data. Just a few more conditions to check:
*
* 1) Make sure thet there is buffer space to receive additional data
* (state.rf_buflen > 0). This could be zero, for example, if read-ahead
* buffering was enabled and we filled the user buffer with data from
* the read-ahead buffers. Aand
* 2) if read-ahead buffering is enabled (CONFIG_NET_NTCP_READAHEAD_BUFFERS > 0)
* and delay logic is disabled (CONFIG_NET_TCP_RECVDELAY == 0), then we
* not want to wait if we already obtained some data from the read-ahead
* buffer. In that case, return now with what we have.
*/
#if CONFIG_NET_TCP_RECVDELAY == 0 && CONFIG_NET_NTCP_READAHEAD_BUFFERS > 0
if (state.rf_recvlen == 0 && state.rf_buflen > 0)
#else
if (state.rf_buflen > 0)
#endif
{
struct uip_conn *conn = (struct uip_conn *)psock->s_conn;