 Relationships |
|
| Anonymous | Login | 2024-05-02 01:42 UTC |
| Main | My View | View Issues | Change Log | Docs |
| Viewing Issue Simple Details [ Jump to Notes ] | [ Issue History ] [ Print ] | ||||||
| ID | Category | Severity | Type | Date Submitted | Last Update | ||
| 0000347 | [1003.1(2008)/Issue 7] Shell and Utilities | Objection | Error | 2010-11-05 12:33 | 2013-04-16 13:06 | ||
| Reporter | geoffclare | View Status | public | ||||
| Assigned To | ajosey | ||||||
| Priority | normal | Resolution | Accepted | ||||
| Status | Closed | ||||||
| Name | Geoff Clare | ||||||
| Organization | The Open Group | ||||||
| User Reference | |||||||
| Section | sh | ||||||
| Page Number | 3167 | ||||||
| Line Number | 105344 | ||||||
| Interp Status | Approved | ||||||
| Final Accepted Text | Note: 0000616 | ||||||
| Summary | 0000347: ASYNCHRONOUS EVENTS on sh page should not be "Default" | ||||||
| Description |
The ASYNCHRONOUS EVENTS section for the sh utility just states "Default". There are several problems with this: 1. It does not take into account the effects of the trap built-in utility (including the special requirements in section 2.11 concerning traps). 2. It conflicts with the requirements for interrupt handling during command line editing in interactive shells, as specified in the EXTENDED DESCRIPTION. 3. It omits other details relating to interactive shells. These are described in the sh man page on Solaris 10 as follows: If the -i flag is present or if the shell input and output are attached to a terminal, this shell is interactive. In this case, TERMINATE is ignored (so that kill 0 does not kill an interactive shell) and INTERRUPT is caught and ignored (so that wait is interruptible). In all cases, QUIT is ignored by the shell. The bash man page is similar, but adds: If job control is in effect, bash ignores SIGTTIN, SIGTTOU, and SIGTSTP. Some experimentation with ksh shows that it also ignores these job control signals. (Both shells seem to ignore them regardless of whether the -m option is set, so I'm not sure what the condition "If job control is in effect" in the bash man page is referring to; perhaps it is whether the shell was compiled with support for job control.) Additionally, where the EXTENDED DESCRIPTION refers to the interrupt character being typed, it should specify the same behaviour on receipt of SIGINT signals generated by other means. The current wording implies that the shell should distinguish between SIGINT signals that are generated by typing the interrupt character and by other means, but in practice shells just handle all SIGINT signals the same way. |
||||||
| Desired Action |
Change the ASYNCHRONOUS EVENTS section from Default. to The sh utility shall take the standard action for all signals (see [xref to Utility Description Defaults]) with the following exceptions: If the shell is interactive, SIGINT signals received during command line editing shall be handled as described in the EXTENDED DESCRIPTION, and SIGINT signals received at other times shall be caught but no action performed. If the shell is interactive, SIGQUIT, SIGTERM, SIGTTIN, SIGTTOU, and SIGTSTP signals shall be ignored. The standard actions, and the actions described above for interactive shells, can be overridden by use of the trap special built-in utility (see [xref to trap] and [xref to 2.11]). At page 3168 line 105387 change Typing the interrupt character in command mode shall cause sh to ... to If sh receives a SIGINT signal in command mode (whether generated by typing the interrupt character or by other means), it shall ... At line 105406 change Terminate command line editing ... to If sh receives a SIGINT signal in insert mode (whether generated by typing the interrupt character or by other means), it shall terminate command line editing ... At page 3178 line 105872 add a new paragraph to RATIONALE In interactive shells, SIGTERM is ignored so that kill 0 does not kill the shell, and SIGINT is caught so that wait is interruptible. |
||||||
| Tags | tc1-2008 | ||||||
| Attached Files | |||||||
|
|
|||||||
|
Relationships |
|
|
Notes |
|
|
(0000616) msbrown (manager) 2010-11-18 16:30 edited on: 2011-01-27 17:32 |
Interpretation response ------------------------ The standard states that the "default" behavior is required here, and conforming implementations must conform to this. However, concerns have been raised about this which are being referred to the sponsor. Rationale: ------------- See the "Description" section above. Notes to the Editor (not part of this interpretation): ------------------------------------------------------- Change the ASYNCHRONOUS EVENTS section from Default. to The sh utility shall take the standard action for all signals (see [xref to Utility Description Defaults]) with the following exceptions: If the shell is interactive, SIGINT signals received during command line editing shall be handled as described in the EXTENDED DESCRIPTION, and SIGINT signals received at other times shall be caught but no action performed. If the shell is interactive: * SIGQUIT and SIGTERM signals shall be ignored. * If the -m option is in effect, SIGTTIN, SIGTTOU, and SIGTSTP signals shall be ignored. * If the -m option is not in effect, it is unspecified whether SIGTTIN, SIGTTOU, and SIGTSTP signals are ignored, set to the default action, or caught. If they are caught the shell shall, in the signal-catching function, set the signal to the default action and raise the signal (after taking any appropriate steps such as restoring terminal settings). The standard actions, and the actions described above for interactive shells, can be overridden by use of the trap special built-in utility (see [xref to trap] and [xref to 2.11]). At page 3168 line 105387 change Typing the interrupt character in command mode shall cause sh to ... to If sh receives a SIGINT signal in command mode (whether generated by typing the interrupt character or by other means), it shall ... At line 105406 change Terminate command line editing ... to If sh receives a SIGINT signal in insert mode (whether generated by typing the interrupt character or by other means), it shall terminate command line editing ... At page 3178 line 105872 add a new paragraph to RATIONALE In interactive shells, SIGTERM is ignored so that kill 0 does not kill the shell, and SIGINT is caught so that wait is interruptible. If the shell does not ignore SIGTTIN, SIGTTOU, and SIGTSTP signals when it is interactive and the -m option is not in effect, these signals suspend the shell if it is not a session leader. If it is a session leader, the signals are discarded if they would stop the process, as required by [xref to XSH 2.4.3] for orphaned process groups. |
|
(0000636) ajosey (manager) 2010-12-16 16:13 |
Comments are due on this interpretation by January 16 2011 |
|
(0000642) jilles (reporter) 2010-12-29 14:40 |
Interactive shells with job control disabled (like set +m) do not really need to ignore SIGTTIN, SIGTTOU and SIGTSTP signals. Not ignoring these signals allows suspending a shell without job control via ctrl+z or similar and allows a better implementation of the non-standard suspend utility than ksh93's kill -s STOP $$ as follows case $- in *m*) set +m; kill -s TSTP 0; set -m;; *) kill -s TSTP 0;; esac which properly suspends any other processes in the process group (otherwise, an outer job control shell may wait in vain for the other processes to stop or terminate). Some ash derivatives such as dash and FreeBSD's sh do not ignore these signals. They also honor +m on the initial command line, unlike bash and ksh93 which enable job control anyway (which is not a disaster, but not what the user asked for). Interactive shells with job control enabled must ignore SIGTTIN, SIGTTOU and SIGTSTP; otherwise, things like tcsetpgrp() may fail. dash and FreeBSD's sh do this. |
|
(0000661) ajosey (manager) 2011-01-27 17:40 |
The notes to the editor in Note: 0000616 have been updated to address the comments made in Note: 0000642. Comments on this revised interpretation are due by Feb 27th. |
|
(0000687) ajosey (manager) 2011-03-04 12:26 |
This is now an approved interpretation |
|
Issue History |
|||
| Date Modified | Username | Field | Change |
| 2010-11-05 12:33 | geoffclare | New Issue | |
| 2010-11-05 12:33 | geoffclare | Status | New => Under Review |
| 2010-11-05 12:33 | geoffclare | Assigned To | => ajosey |
| 2010-11-05 12:33 | geoffclare | Name | => Geoff Clare |
| 2010-11-05 12:33 | geoffclare | Organization | => The Open Group |
| 2010-11-05 12:33 | geoffclare | Section | => sh |
| 2010-11-05 12:33 | geoffclare | Page Number | => 3167 |
| 2010-11-05 12:33 | geoffclare | Line Number | => 105344 |
| 2010-11-05 12:33 | geoffclare | Interp Status | => --- |
| 2010-11-18 16:27 | msbrown | Tag Attached: tc1-2008 | |
| 2010-11-18 16:30 | msbrown | Interp Status | --- => Pending |
| 2010-11-18 16:30 | msbrown | Note Added: 0000616 | |
| 2010-11-18 16:30 | msbrown | Status | Under Review => Interpretation Required |
| 2010-11-18 16:30 | msbrown | Resolution | Open => Accepted |
| 2010-11-18 16:30 | msbrown | Description Updated | |
| 2010-11-18 16:30 | msbrown | Desired Action Updated | |
| 2010-11-18 16:31 | msbrown | Final Accepted Text | => Note: 0000616 |
| 2010-12-16 16:13 | ajosey | Interp Status | Pending => Proposed |
| 2010-12-16 16:13 | ajosey | Note Added: 0000636 | |
| 2010-12-29 14:40 | jilles | Note Added: 0000642 | |
| 2011-01-27 17:32 | geoffclare | Note Edited: 0000616 | |
| 2011-01-27 17:40 | ajosey | Note Added: 0000661 | |
| 2011-03-04 12:26 | ajosey | Interp Status | Proposed => Approved |
| 2011-03-04 12:26 | ajosey | Note Added: 0000687 | |
| 2013-04-16 13:06 | ajosey | Status | Interpretation Required => Closed |
| Mantis 1.1.6[^] Copyright © 2000 - 2008 Mantis Group |  |
