#223 SecurityManager at interpreter level

jfaucher — Sun, 16 Nov 2025 07:17:40 -0000

Small rework for NativeActivation and NativeCode:
- added getEffectiveSecurityManager
- don't mix getSecurityManager and getEffectiveSecurityManager

Still a WIP because the security of the setSecurityManager BIF has not yet been taken into account.

Attached: SecurityManager-v3.patch (replace the previous patch)

#223 SecurityManager at interpreter level

jfaucher — Sat, 15 Nov 2025 19:05:02 -0000

Attachments has changed:

Diff:

--- old
+++ new
@@ -1 +1,2 @@
+SecurityManager-v2.patch (9.6 kB; application/octet-stream)
 interpreter.zip (10.4 kB; application/zip)

SecurityManager at interpreter level

jfaucher — Sat, 15 Nov 2025 19:03:58 -0000

Creating this ticket because devel-list seems to reject emails with attached zip.

Still a WIP.

Preliminary note:
The setSecurityManager methods are protected, allowing a security manager to prohibit its own replacement by the monitored agent program.
The new setSecurityManager BIF is not protected.

The setSecurityManager BIF must be replaced by something that can be monitored.
Could be a Sys function.
Could be a method, but where?

Or could call directly the checkFunctionCall of the security manager from the setSecurityManager BIF.

On 14 Nov 2025, at 23:18, Rony G. Flatscher Rony.Flatscher@wu.ac.at wrote:
Just to see whether I understood what this means:

If one implements the above BIF and sets a security manager for an interpreter instance, then all code executed by that instance will be supervised with that security manager?

yes but additional changes are needed.

Would this also include any called (not required) program?

yes, with the changes below.

SM = SecurityManager
An SM is a wrapper of a manager, a manager is the RexxObject you pass to setSecurityManager.

Some classes like Activity, RexxActivation and PackageClass have a securityManager (an attribute).
Some classes like MethodClass and RoutineClass delegate to their Code object.
The Code object delegates to its PackageClass.

Activity and RexxActivation have an effective SecurityManager (a method).
The effective SM returns the stored SM if it is not NULL, otherwise the interpreter's SM.

Changes applied:
Added an effective SM on PackageClass: getEffectiveSecurityManager()
Replaced the use of getSecurityManager() by the new getEffectiveSecurityManager() of PackageClass.
Modified the implementation of the effective SM to test if the stored SM is not NULL and active.
An active SM is an SM with a non NULL manager attribute.

If one sets the security manager to the interpreter it may be the case that the LOCAL, ENVIRONMENT and REQUIRES checkpoints do not get called. However, if one sets a security manager to the routine object (result of .routine~newFile(...)), then these checkpoints get called by sending the appropriate messages to the security manager?

The REQUIRES checkpoints are called, that was the objective of this WIP.
I made minor changes to have the LOCAL and ENVIRONMENT called:
Use the new getEffectiveSecurityManager() of PackageClass, instead of getSecurityManager().

So to get a fully functional security manager one needs to do two things, a) set the interpreter instance to a security manager, and b) set the routine object to a security manager?

b) is no longer needed.

It will be possible to remove the security manager from the instance by invoking the BIF without an argument?

yes

If one does this, then the requires, local and environment checkpoints do not get sent anymore?

yes, if there are not more specialized SMs (on routine or method or package).
The interpreter SM is just a default SM, it never replaces more specialized SMs.

Remember, a ticket was already opened by JMB:
https://sourceforge.net/p/oorexx/bugs/1886
Thank you for pointing that out!

Other older tickets:
https://sourceforge.net/p/oorexx/bugs/1422/
https://sourceforge.net/p/oorexx/bugs/1487/

Patch attached: SecurityManager-v2.patch
Test attached: interpreter.zip

SecurityManager at interpreter level

jfaucher — Sat, 15 Nov 2025 19:03:58 -0000

Ticket 223 has been modified: SecurityManager at interpreter level
Edited By: jfaucher (jfaucher)

#221 Windows Clipboard Unicode contents

Rony G. Flatscher — Fri, 02 May 2025 10:31:08 -0000

status: pending --> closed

#212 Change name of samples readme into readme.txt

Per Olov Jonsson — Mon, 10 Mar 2025 22:02:46 -0000

status: open --> closed
assigned_to: Per Olov Jonsson

#212 Change name of samples readme into readme.txt

Per Olov Jonsson — Mon, 10 Mar 2025 21:03:08 -0000

0ReadMe.first is for Windows, this patch was for readme in the samples directory

#181 Suggested patch: Update init script to LSB standard

Erich — Sat, 13 Jan 2024 18:20:22 -0000

status: open --> wont-fix

#217 test

Erich — Sat, 13 Jan 2024 18:17:48 -0000

status: open --> wont-fix

#213 oorexx on the BSD:s

Erich — Sat, 13 Jan 2024 18:17:04 -0000

status: open --> closed

Recent changes to patches

#223 SecurityManager at interpreter level

#223 SecurityManager at interpreter level

SecurityManager at interpreter level

SecurityManager at interpreter level

#221 Windows Clipboard Unicode contents

#212 Change name of samples readme into readme.txt

#212 Change name of samples readme into readme.txt

#181 Suggested patch: Update init script to LSB standard

#217 test

#213 oorexx on the BSD:s