Re: [PATCH]fix VMX "xm console" issue

[Anthony Liguori <aliguori@us.ibm.com>]

Yu, Ping Y wrote:

>>Looking at the qemu code to create a PTY device, it's not quite optimal.
>>
>>For instance, it's not suggested that you pass a name parameter to
>>openpty() b/c it's unclear what's a safe size of the buffer.  Instead,
>>ptsname() should be called on the slave fd.
>>    
>>
>This is true. Maybe a small patch is needed. 
>
>  
>
>>I think the right fix is to tcsetattr() a cfmakeraw() on the slave PTY
>>fd after openpty() is called in vl.c.  Otherwise, it's quite possible
>>(and likely) to get strange buffering behavior depending on the terminal
>>attributes that are inherited when ioemu is spawned.
>>    
>>
>You are right. Currently, there is no control for VMX pty. Following your direction 
>I made a patch and it sounds OK in testing.
>
>diff -r eb1169f92d81 tools/ioemu/vl.c
>--- a/tools/ioemu/vl.c  Sun Nov 27 13:09:46 2005
>+++ b/tools/ioemu/vl.c  Wed Nov 30 19:16:34 2005
>@@ -1218,15 +1218,20 @@
> #if defined(__linux__)
> CharDriverState *qemu_chr_open_pty(void)
> {
>-    char slave_name[1024];
>+    char *slave_name;
>     int master_fd, slave_fd;
>+    struct termios term;
>
>     /* Not satisfying */
>-    if (openpty(&master_fd, &slave_fd, slave_name, NULL, NULL) < 0) {
>+    if (openpty(&master_fd, &slave_fd, NULL, NULL, NULL) < 0) {
>         return NULL;
>     }
>     fprintf(stderr, "char device redirected to %s\n", slave_name);
>-    store_console_dev(domid, slave_name);
>+
>+    /* set attritrute */
>+    cfmakeraw(&term);
>+    tcsetattr(slave_fd, TCSAFLUSH, &term);
>+    store_console_dev(domid, ptsname(slave_fd));
>     return qemu_chr_open_fd(master_fd, master_fd);
> }
> #else
>  
>
Excellent!  Looks good to me.

Regards,

Anthony Liguori