Antti Antinoja

Thank you. Below some details. This is a Genoo box with following related software and libraries: ``` $ equery l gcc xfce4-session binutils openrc glib * Searching for gcc ......

> What program is your drop target? `xfce-base/thunar-4.16.11`

Applied three patches.. ``` c994f3e6e1b1af4599d34147139e9012e3941e35c496ca7d6c7eddf8e1b09d14 0006.patch afc81a8e933d888c455e3da41824f999f95a93bfb83a5abdd105b9512a94c062 0008.patch 868084b35c38bde863c83f0abd41ec49b315acce087ba17beb728dd10364ba2c 0009.patch ``` .. on top of 63dad3ceb7a3d1dc2288a0df6b054a6ea244a8b6. Could not reproduce either one of the issues with the newly compiled executable. Looks...

Very good. Thank you for handling these issues.

Ok. Yes. By reading ``man mrecv`` one could assume that sz would be set to "-1" and errno to ``EMSGSIZE: The data won't fit into the supplied buffer.``.

Thank you!

Just for curiosity, would this be an acceptable way to implement EMSGSIZE for mrecv? ```diff index c493bba..fea5901 100644 --- a/msock.c +++ b/msock.c @@ -42,7 +42,12 @@ ssize_t dill_mrecv(int s, void...

Altered step6.c: ```c coroutine void dialogue(int s, int ch) { int op = CONN_ESTABLISHED; int rc = chsend(ch, &op, sizeof(op), -1); assert(rc == 0); int64_t deadline = now() + 60000;...

Reverted msock.c back to original. ``dill_suffix_mrecvl`` seems to have this while loop: ```c 164 while(first->iol_len == 0) { 165 if(!it.iol_next) {self->inerr = 1; errno = EMSGSIZE; return -1;} 166 it...

Ok. I think there are two strategies on how the input buffering was dealt with (via a temporary buffer or directly to the buffer provided by the user).