Hi RP,
On 2/26/21 5:51 AM, Richard Purdie wrote:
On Thu, 2021-02-25 at 03:25 -0800, Robert Yang wrote:
In many scenarios the receipt of an EOF will coincide with the
teardown of the file which the FD is associated with by the
application on the other side of the file pipe. If a timeout (-t) is
used socat continues to poll for new data (read) and send any new data
(write) until the timeout. This can lead to input/output errors as the
FD exists still but the file it was openned on is gone, and no check
is done on writes.
To get around this a check is added after the EOF is received that the
file still exists, and if not we proceed to exit immediately. This
will avoid i/o errors and will also cause a quicker return as we don't
have to wait for the timeout.
The original author is Mark Asselstine <[email protected]>.
Signed-off-by: Robert Yang <[email protected]>
---
...le-associated-with-an-FD-is-destroye.patch | 62 +++++++++++++++++++
.../socat/socat_1.7.4.1.bb | 1 +
2 files changed, 63 insertions(+)
create mode 100644
meta/recipes-connectivity/socat/files/Exit-when-the-file-associated-with-an-FD-is-destroye.patch
diff --git
a/meta/recipes-connectivity/socat/files/Exit-when-the-file-associated-with-an-FD-is-destroye.patch
b/meta/recipes-connectivity/socat/files/Exit-when-the-file-associated-with-an-FD-is-destroye.patch
new file mode 100644
index 00000000000..b5ac62f6169
--- /dev/null
+++
b/meta/recipes-connectivity/socat/files/Exit-when-the-file-associated-with-an-FD-is-destroye.patch
@@ -0,0 +1,62 @@
+From 956d4b00a101f61b4cfaafa0856b646646536492 Mon Sep 17 00:00:00 2001
+From: Mark Asselstine <[email protected]>
+Date: Mon, 7 Dec 2015 21:27:35 -0500
+Subject: [PATCH] Exit when the file associated with an FD is destroyed
+
+In many scenarios the receipt of an EOF will coincide with the
+teardown of the file which the FD is associated with by the
+application on the other side of the file pipe. If a timeout (-t) is
+used socat continues to poll for new data (read) and send any new data
+(write) until the timeout. This can lead to input/output errors as the
+FD exists still but the file it was openned on is gone, and no check
+is done on writes.
+
+To get around this a check is added after the EOF is received that the
+file still exists, and if not we proceed to exit immediately. This
+will avoid i/o errors and will also cause a quicker return as we don't
+have to wait for the timeout.
+
+Signed-off-by: Mark Asselstine <[email protected]>
+
+Upstream-Status: Submitted [[email protected]]
+
+Signed-off-by: Robert Yang <[email protected]>
This was written in 2015, when was it submitted? What was the response?
An idea of the current status would be good for a patch of this age.
I sent it to upstream yesterday, I got a reply from upstream today which asked
for a testcase, let's hold on this patch and wait for upstream to merge it.
// Robert
Cheers,
Richard
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#148637):
https://lists.openembedded.org/g/openembedded-core/message/148637
Mute This Topic: https://lists.openembedded.org/mt/80899603/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
