|
2017-05-24
| ||
| 08:43 | • Ticket [19a8c9399d] var-22.1 failing on trunk with TCL_MEM_DEBUG status still Closed with 5 other changes artifact: 3c7a68984b user: jan.nijtmans | |
|
2017-05-23
| ||
| 16:00 | • Closed ticket [19a8c9399d]. artifact: e3eed34448 user: dgp | |
| 16:00 | [19a8c9399d] Plug mem leak in TIP 459 machinery. check-in: f1c99ff14c user: dgp tags: trunk | |
| 15:37 | • Ticket [19a8c9399d] var-22.1 failing on trunk with TCL_MEM_DEBUG status still Open with 3 other changes artifact: 38b6568dae user: dgp | |
|
2017-05-22
| ||
| 09:51 | • Ticket [19a8c9399d]: 4 changes artifact: eac72f507e user: jan.nijtmans | |
|
2017-05-21
| ||
| 15:23 | • Ticket [19a8c9399d]: 3 changes artifact: b0c7432fa2 user: aspect | |
| 15:22 | • New ticket [19a8c9399d]. artifact: 7b69c9e3bc user: aspect | |
| Ticket UUID: | 19a8c9399d025831a15d5bd0047a005c799e844e | |||
| Title: | var-22.1 failing on trunk with TCL_MEM_DEBUG | |||
| Type: | Bug | Version: | trunk | |
| Submitter: | aspect | Created on: | 2017-05-21 15:22:58 | |
| Subsystem: | 47. Bytecode Compiler | Assigned To: | dgp | |
| Priority: | 5 Medium | Severity: | Minor | |
| Status: | Closed | Last Modified: | 2017-05-24 08:43:02 | |
| Resolution: | Fixed | Closed By: | jan.nijtmans | |
| Closed on: | 2017-05-24 08:43:02 | |||
| Description: |
With --enable-symbols=all (I presume TCL_MEM_DEBUG is the trigger), checkin [abcf4db5930f] introduced some test failures, of which var-22.1 seems to point at the root. | |||
| User Comments: |
jan.nijtmans added on 2017-05-24 08:43:02:
Thanks Don, for fixing this!!! dgp added on 2017-05-23 16:00:32: Looks like a simple lack of a required ckfree(). Fixed in trunk. Thank you for reporting the problem! dgp added on 2017-05-23 15:37:22: It appears that the TIP 459 implementation created a memory leak in a cycle of interp creation and deletion. jan.nijtmans added on 2017-05-22 09:51:36: Thanks for the report! Testcase var-22.1 is a test for a memory leak. Commit [abcf4db5930f] starts keeping track of files loaded during package initialization in a Tcl_Obj* list. I suspect that the test-case sees this as a leak, even though in reality it isn't. For sure, the test-case needs to be adapted some way to correct this, or maybe I'm wrong and there is a leak somehow. But anyway, I don't think this can be considered critical. I will investigate further. | |||