[asterisk-bugs] [JIRA] (ASTERISK-22351) Segfault in LIBEDIT_INTERNAL after tgetstr()

[Walter Doekes (JIRA)]

    [ https://issues.asterisk.org/jira/browse/ASTERISK-22351?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=209516#comment-209516 ] 

Walter Doekes commented on ASTERISK-22351:
------------------------------------------

{quote}
in the line you mention, cap is the result of calling tgetstr (the call the originate the failure). It's clear that the segfault is originated by this:
{quote}

Well, no.

The backtrace you pasted shows that the segfault happens upon touching the *result* of the tgetstr() call. This: {{0xffffffffffffca60 <Address 0xffffffffffffca60 out of bounds>}}
It's not the call itself that segfaults.

If {{( &buf <= cap < &buf+TC_BUFSIZE )}} then there shouldn't be a problem, since cap is in a valid range.

{quote}
By the way, where is tgetstr function?
{quote}
It fetches the "right" console_code to perform a specific non-printing action on the tty:
{noformat}
{ "al", "add new blank line" },
{noformat}

Over here, it returns:
{noformat}
"\x1b[L" (try: printf '\x1b[L')
{noformat}
Or:
{noformat}
{ "cl", "clear screen" },
{noformat}
Returns:
{noformat}
"\x1b[H\x1b[2J" (try: printf '\x1b[H\x1b[2J')
{noformat}
                
> Segfault in LIBEDIT_INTERNAL after tgetstr()
> --------------------------------------------
>
>                 Key: ASTERISK-22351
>                 URL: https://issues.asterisk.org/jira/browse/ASTERISK-22351
>             Project: Asterisk
>          Issue Type: Bug
>      Security Level: None
>    Affects Versions: 1.8.23.0, 11.5.0
>         Environment: HP Proliant DL320e G8
> Debian Wheezy (kernel 3.2.0-4-amd64)
> DAHDI 2.7.0 (no cards installed)
> Libpri 1.4.14
>            Reporter: A. Iglesias
>         Attachments: issueA22351_larger_buf_and_offset.patch, issueA22351_larger_buf.patch
>
>
> New installation in this server. After everything is installed and asterisk started, anytime a try to open an asterisk console with asterisk -vvvvvvvvvr I get a "Segmentation fault" and console doesn't open, but asterisk keeps running. If I stop asterisk and try to start it with asterisk -vvvvvvc, it fails in the same way.
> I've been doing some research, and found a guy with the same problem in the past, related whit the value of the TERM system variable. In my case is "xterm", but if I change it for a non existent value ('lalalala' for example) then console starts.
> Digging deeper, I've found that the problem is in main/editline/term.c , in line 960 when executing this:
> term_alloc(el, t, tgetstr((char *)t->name, &area));
> If I change this line with the one executed when TERM has an unknown value...
> term_alloc(el, t, NULL);
> ... and recompile, then console starts.
>  the problem seems to be in the tgetstr function, because if I add a line just with this...
> tgetstr((char *)t->name, &area);
> ... I get the same error.
> I'll try to get core dump stuff, but maybe with this info is enough at the moment to find a solution. I've been trying to find the problem/solution studying the code, but no luck yet.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.asterisk.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira