Fossil Forum

segfault trying to click on manifest link
Login

segfault trying to click on manifest link

segfault trying to click on manifest link

(1) By anonymous on 2020-08-16 15:31:54 [link] [source]

Hi, I was trying to look at manifests from this repo using the builtin links for each checkin and for each link that I try it gives me this error (the numbers vary from checkin to checkin):

Segfault (0) [0x465879] (1) [0x734a50] (2) [0x756f25] (3) [0x448484] (4) [0x45f7bd] (5) [0x467a31] (6) [0x4686de] (7) [0x466890] (8) [0x405b3c] (9) [0x727776] (10) [0x727d65] (11) [0x405e09]

I don't know what is the cause of this, I didn't have time to download the new 2.12 released today and try it on my repos.

(2.1) By andygoth on 2020-08-16 15:58:50 edited from 2.0 in reply to 1 [source]

I'm seeing the same exact thing too right now and am debugging. Actually I have to run right this second, so here's some backtrace:

0x00007ffff76f0915 in __strrchr_sse2 () from /lib64/libc.so.6
(gdb) ba
#0  0x00007ffff76f0915 in __strrchr_sse2 () from /lib64/libc.so.6
#1  0x0000000000458524 in file_extension (zFileName=0x0) at bld/file_.c:2408
#2  0x0000000000476a61 in artifact_page () at bld/info_.c:2456
#3  0x0000000000480e71 in process_one_web_page (zNotFound=0x0, pFileGlob=0x0, 
    allowRepoList=0) at bld/main_.c:1961
#4  0x000000000048201c in cmd_http () at bld/main_.c:2591
#5  0x000000000047f2d2 in fossil_main (argc=2, argv=0x7fffffffdfa8)
    at bld/main_.c:938
#6  0x000000000047ea0a in main (argc=2, argv=0x7fffffffdfa8) at bld/main_.c:648
(gdb) f 1
#1  0x0000000000458524 in file_extension (zFileName=0x0) at bld/file_.c:2408
2408	  const char * zExt = strrchr(zFileName, '.');
(gdb) up 
#2  0x0000000000476a61 in artifact_page () at bld/info_.c:2456
2456	        zExt = file_extension(zFileName);
(gdb) up
#3  0x0000000000480e71 in process_one_web_page (zNotFound=0x0, pFileGlob=0x0, 
    allowRepoList=0) at bld/main_.c:1961
1961	          pCmd->xFunc();
(gdb) up
#4  0x000000000048201c in cmd_http () at bld/main_.c:2591
2591	  process_one_web_page(zNotFound, glob_create(zFileGlob), allowRepoList);

(3) By Richard Hipp (drh) on 2020-08-16 16:02:38 in reply to 1 [link] [source]

To repro this bug, first create a text file (named "r.txt") that contains:

   GET /artifact/4368f52961d0bc57 HTTP/1.0

Then run:

   fossil test-http <r.txt

Using the technique above, it is easy to bisect. The problem check-in appears to be b699040d701464ce, which is the merge of the new line-numbering enhancements.

(4) By Stephan Beal (stephan) on 2020-08-16 16:12:56 in reply to 3 [link] [source]

The problem check-in appears to be b699040d701464ce, which is the merge of the new line-numbering enhancements.

That wasn't really intended to be merged for 2.12 :/, but i will look at it right this minute.

Be aware that there's currently a mess on the trunk right now, as describe in /forumpost/f5b1b55b65, and i don't want to check anything in until that's resolved.

(5) By Stephan Beal (stephan) on 2020-08-16 16:18:23 in reply to 4 [link] [source]

, but i will look at it right this minute.

i see that that's resolved. Facepalm to me for that one and thank you for spotting the error so quickly.

(6) By Richard Hipp (drh) on 2020-08-16 16:28:11 in reply to 4 [link] [source]

Be aware that there's currently a mess on the trunk right now

Sorry - I thought this was ready to merge. But in any event, the merge occurred after the 2.12 release, so the mess, and the bug, are not found in 2.12.

(7) By sean (jungleboogie) on 2020-08-16 16:33:42 in reply to 6 [link] [source]

But in any event, the merge occurred after the 2.12 release

I do see you made a release...nice.

Please consider updating index.wiki to reflect this.

Also why isn't the tag colored the same as other release tags? https://fossil-scm.org/fossil/timeline?t=release

(8) By Stephan Beal (stephan) on 2020-08-16 16:36:08 in reply to 6 [link] [source]

Sorry - I thought this was ready to merge.

It wasn't not ready - it was in a usable state aside, of course, from that one NULL pointer :|. i misunderstood, though:

But in any event, the merge occurred after the 2.12 release...

Oh, indeed!!! In my panic i misread the timeline and thought you merged that just before tagging 2.12. That's a relief.