Not related to the post I'm now replying to, so don't try to comment link this one. 8-).

Tommi,

I've noticed that recently, every time you check in to my server, the server breaks down. This reminds me of one of the first bugs that Johan and I discovered when beta testing JamNNTPd, but then it was a non-existing Jam folder that made JamNNTPd exit without a trace.

Recently I've had this again, but this time it was a corrupt Jam base that was the culprit. Deleting that specific base solved the problem then.

Now I ask you, do you remember what exact group you were trying to access here when you noticed that it all froze? I'm investigating this seriously, and the next time you freeze my server, I think I can find out what happened. Thanks in advance,



..

--- Mozilla/5.0 (Windows; U; Windows NT 5.1; sv-SE; rv:1.9.1.16) Gecko/20101125
* Origin: news://eljaco.se (2:203/2)

13 Feb 21 12:11, Björn Felten wrote to Tommi Koivula:

Not related to the post I'm now replying to, so don't try to comment link this one. 8-).

Tommi,

I've noticed that recently, every time you check in to my server, the server breaks down.
This reminds me of one of the first bugs that Johan and I discovered when beta testing
JamNNTPd, but then it was a non-existing Jam folder that made JamNNTPd exit without a trace.

Recently I've had this again, but this time it was a corrupt Jam base that was the culprit.
Deleting that specific base solved the problem then.

Now I ask you, do you remember what exact group you were trying to access here when you
noticed that it all froze? I'm investigating this seriously, and the next time you freeze my
server, I think I can find out what happened. Thanks in advance,

Bummer.

Let's see if I can crash your server again.

'Tommi

---
* Origin: rbb.fidonet.fi (2:221/360)

Bj÷rn Felten <0@2.203.2> wrote:

Not related to the post I'm now replying to, so don't try to
comment link this one. 8-).

Tommi,

I've noticed that recently, every time you check in to my
server, the server breaks down. This reminds me of one of the first
bugs that Johan and I discovered when beta testing JamNNTPd, but
then it was a non-existing Jam folder that made JamNNTPd exit
without a trace.

Recently I've had this again, but this time it was a corrupt Jam
base that was the culprit. Deleting that specific base solved the
problem then.

Now I ask you, do you remember what exact group you were trying
to access here when you noticed that it all froze? I'm
investigating this seriously, and the next time you freeze my
server, I think I can find out what happened. Thanks in advance,

There!

[18:11:52] NNTP> GROUP FIDONEWS_1
[18:11:52] NNTP< 211 16000 0 15999 FIDONEWS_1 Group selected
* getting xover 15700 - 15999 in FIDONEWS_1...
[18:11:52] NNTP> XOVER 15700-15999
[18:11:52] NNTP< 224 Overview information follows
** LibSylph-WARNING: fd_recv(): failed with 10054 (errno = 0)

** LibSylph-WARNING: [18:11:52] error occurred while getting xover.

** error occurred while getting xover.
* creating NNTP connection to eljaco.se:4119 ...
** LibSylph-WARNING: [18:11:55] Can't connect to NNTP server:
eljaco.se:4119

** Can't connect to NNTP server: eljaco.se:4119

--- Sylpheed 3.7.0 (GTK+ 2.24.30; i686-pc-mingw32)
* Origin: nntps://news.fidonet.fi - Lake Ylo - Finland (2:221/6.0)

Björn wrote (2021-02-13):

Not related to the post I'm now replying to, so don't try to comment link this one. 8-).

Tommi,

I've noticed that recently, every time you check in to my server, the server breaks down. This reminds me of one of the first bugs that Johan and I discovered when beta testing JamNNTPd, but then it was a non-existing Jam folder that made JamNNTPd exit without a trace.

I crashed JamNNTPd recently when I tried to fetch a newsgroup with Soupgate.. Unfortunately I don't know why it crashed, nothing in the logs and maybe unrelated to the crashs you are seeing.

---
* Origin: . (2:280/464.47)

Oli wrote:

I crashed JamNNTPd recently when I tried to fetch a newsgroup with
Soupgate.. Unfortunately I don't know why it crashed, nothing in the
logs and maybe unrelated to the crashs you are seeing.

I have seen JamNNTPd to crash when client (hotdoged) tried to access a
message that was deleted.

BTW. Are you able to compile Spoon correctly in Linux? I can compile it
but it crashes when sending messages. Works but crashes.

'Tommi

--- Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:60.0) Gecko/20100101 SeaMonkey/2.53.6
* Origin: - rbb.fidonet.fi - Lake Ylo - Finland - (2:221/1)

Tommi wrote (2021-02-13):

Oli wrote:

I crashed JamNNTPd recently when I tried to fetch a newsgroup with
Soupgate.. Unfortunately I don't know why it crashed, nothing in the
logs and maybe unrelated to the crashs you are seeing.

I have seen JamNNTPd to crash when client (hotdoged) tried to access a message that was deleted.

BTW. Are you able to compile Spoon correctly in Linux? I can compile it but it crashes when sending messages. Works but crashes.

My mistake, of course it was Spoon not Soupgate. Does it crash with some "double free" error? I don't think anyone has found a fix for it. I assume the problem is somewhere hidden in socket.cpp (there are a couple of deletes). Maybe when the connection closes and maybe harmless, because everything's done already. I also gave up on Soupgate, it's the chainsaw massacre of FTN<->RFC822 gating.

---
* Origin: . (2:280/464.47)

Saturday February 13 2021 22:34, Oli wrote to Tommi Koivula:

BTW. Are you able to compile Spoon correctly in Linux? I can compile it
but it crashes when sending messages. Works but crashes.

My mistake, of course it was Spoon not Soupgate. Does it crash with some "double free" error?

Yes, that's it.

===
Posting news to 192.168.1.24...
+ Connecting to NNTP host 192.168.1.24...
* Posting article to rbb.test
+ Signing off...
free(): double free detected in tcache 2
===

I don't think anyone has found a fix for it. I assume the problem is somewhere hidden in socket.cpp (there are a couple of deletes).
Maybe when the connection closes and maybe harmless, because everything's done
already. I also gave up on Soupgate, it's the chainsaw massacre of FTN<->RFC822
gating.

Well, it still runs fine in OS/2, so no worries. :)

'Tommi

---
* Origin: rbb.fidonet.fi (2:221/360)

There!

[18:11:52] NNTP> GROUP FIDONEWS_1
[18:11:52] NNTP< 211 16000 0 15999 FIDONEWS_1 Group selected

Thanks a million!

Now, if only I had a tool to correct a corrupt JAM-base... :(



..

--- Mozilla/5.0 (Windows; U; Windows NT 5.1; sv-SE; rv:1.9.1.16) Gecko/20101125
* Origin: news://eljaco.se (2:203/2)

Hi! Björn,


On 14 Feb 2021, Björn Felten said the following...

[18:11:52] NNTP> GROUP FIDONEWS_1
[18:11:52] NNTP< 211 16000 0 15999 FIDONEWS_1 Group selected

Thanks a million!
Now, if only I had a tool to correct a corrupt JAM-base... :(

Replace with yesterday's (or whenever) backup and %rescan from your favourite link. [shrug]

Cheers,
Paul.

--- Mystic BBS v1.12 A46 2020/08/26 (Linux/32)
* Origin: Quinn's Rock - stuck in another Linux VM, again! (3:640/1384)

TK> There!

TK> [18:11:52] NNTP> GROUP FIDONEWS_1
TK> [18:11:52] NNTP< 211 16000 0 15999 FIDONEWS_1 Group selected

Thanks a million!

Now, if only I had a tool to correct a corrupt JAM-base... :(

Isn't there any tools in crashmail to do that?

GoldED is a good start.

--- Sylpheed 3.7.0 (GTK+ 2.24.30; i686-pc-mingw32)
* Origin: news://eljaco.se (2:203/2)

Tommi wrote (2021-02-14):

Now, if only I had a tool to correct a corrupt JAM-base... :(

Isn't there any tools in crashmail to do that?

I don't see one, only crashmaint for packing. I guess hpt has one.

GoldED is a good start.

does Golded repair a message base?

---
* Origin: . (2:280/464.47)

Hi Oli.

14 Feb 21 08:31:46, you wrote to me:

Now, if only I had a tool to correct a corrupt JAM-base... :(

Isn't there any tools in crashmail to do that?

I don't see one, only crashmaint for packing.

Ok.

I guess hpt has one.

Sure it does. I think you could use hptutil with some minimal fidoconfig without actually running hpt as a tosser.

=== Cut ===
Usage: hptutil [options]
Options: sort - sort unread messages by time and date
link - reply-link messages
purge - purge areas
pack - pack areas
fix - fix base (hptutil fix -? for more help)
-j - link Jam areas by CRC (great speed-up)
-k - keep import.log file
-q - quiet mode (no screen output)
-i <filename> - alternative import.log
=== Cut ===

GoldED is a good start.

does Golded repair a message base?

Not really, but it may show you errors, also it may recover some messages from broken base copying them to a new base.

'Tommi

---
* Origin: IPv6 Point at [2001:470:1f15:cb0:2:221:1:1] (2:221/1.1)

13 Feb 2021 22:45, you wrote to Oli:

I have seen JamNNTPd to crash when client (hotdoged) tried to access a message that was deleted.

I also had this issue. This fix seems to work:

https://github.com/cnb/jamnntpd/commit/3e075c2a904cc8ca66a2753543db4c8d6308a457

Carlos

--- GoldED+/W32-MSVC 1.1.5-b20180707
* Origin: cyberiada (2:341/234.1)

Hello, Carlos Navarro.
On 09/04/2026 13.07 you wrote:

13 Feb 2021 22:45, you wrote to Oli:

I have seen JamNNTPd to crash when client (hotdoged) tried to access a
message that was deleted.

I also had this issue. This fix seems to work: https://github.com/cnb/jamnntpd/commit/3e075c2a904cc8ca66a2753543db4c8d6308a457

I just applied the fix. Lets see. ;)

Thanks!

Carlos

--
Tommi

---
* Origin: news://news.fidonet.fi (2:221/1.0)

On 9.4.2026 17.24, Tommi Koivula wrote:

https://github.com/cnb/jamnntpd/commit/3e075c2a904cc8ca66a2753543db4c8d6308a457

I just applied the fix. Lets see.;)

I deleted some messages, tried to read, no crash yet.

Fixed the OS/2 version too. :)

'Tommi

--- FastEcho/2 1.46.1 Revival
* Origin: nntp://rbb.fidonet.fi - Finland (2:221/360.0)

10/04/2026 7:14, Tommi Koivula -> Carlos Navarro:

I deleted some messages, tried to read, no crash yet.

Fixed the OS/2 version too. :)

Good to know, thanks for reporting.

Carlos

---
* Origin: cyberiada-NNTP (2:341/234.99)

On 11.4.2026 19.58, Carlos Navarro wrote:

I deleted some messages, tried to read, no crash yet.

Ok, I tried to reply to a message that was deleted. So Jamnntpd tried to find the message for reply linking. But it was not there.

Crashed.

Linux jamnntpd.log:

12-Apr-26 16:45:10 (2001:999:41c:1626:11d2:628a:3d43:ea64:56646) Connection established to n4ft960xor07mixg5t0-1.v6.elisa-mobile.fi
12-Apr-26 16:45:10 (2001:999:41c:1626:11d2:628a:3d43:ea64:56646) Logged in as tommi
12-Apr-26 16:45:10 (2001:999:41c:1626:11d2:628a:3d43:ea64:56646) Newsreader : HotdogEd/2.13.5 (Android; Google Android; rv:1) Hotdoged/1772776635000 HotdogEd/2.13.5
12-Apr-26 16:45:10 (2001:999:41c:1626:11d2:628a:3d43:ea64:56646) Accessing JAM message base "/bbs/fmail/msgbase/jam/rbb.test.1"

linux syslog:
Apr 12 16:45:10 kco kernel: [2065949.823079] jamnntpd[4055351]: segfault at 21cd8c ip 000000005656f8ca sp 00000000f7ce4910 error 4 in jamnntpd[138ca,5655c000+1a000] likely on CPU 5 (core 0, socket 5)

'Tommi

---
* Origin: news://news.fidonet.fi (2:221/1.0)

Hello, Tommi!

Sunday April 12 2026 16:54, from Tommi Koivula -> Carlos Navarro:

Ok, I tried to reply to a message that was deleted. So Jamnntpd tried
to find the message for reply linking. But it was not there.
Crashed.

The reproduction steps are very helpful.

linux syslog:
Apr 12 16:45:10 kco kernel: [2065949.823079] jamnntpd[4055351]:
segfault at 21cd8c ip 000000005656f8ca sp 00000000f7ce4910 error 4 in jamnntpd[138ca,5655c000+1a000] likely on CPU 5 (core 0, socket 5)

Doesn't ring a bell. I don't have your binary to check what IP (not IPv4/IPv6 but instruction pointer) it corresponds to.

For better assistance, you should've included the backtrace of the sigsegv. You probably have coredump somewhere around, or use the `ulimit -c unlimited`, cause some distros will mess up with it.

As a last resort, you can just run the jamnntpd from gdb and execute your crash scenario, and gdb will stop at crash point, so you just type `bt` for backtrace and post it here.

Best Regards, Nil
--- GoldED+/LNX 1.1.5-b20260305
* Origin: Gemini can make mistakes, so double-check it (2:5015/46)

Hello, Carlos!

Saturday April 11 2026 18:58, from Carlos Navarro -> Tommi Koivula:

Fixed the OS/2 version too. :)

Good to know, thanks for reporting.

I wonder you guys compile right on OS/2 or in cross-compile env?
GCC-9/GCC-10 or Watcom2 (or maybe even not Open Watcom?), or unofficial GCC 14/15?

Best Regards, Nil
--- GoldED+/LNX 1.1.5-b20260305
* Origin: Gemini can make mistakes, so double-check it (2:5015/46)

12/04/2026 15:54, Tommi Koivula -> Carlos Navarro:

Ok, I tried to reply to a message that was deleted. So Jamnntpd tried to
find the message for reply linking. But it was not there.

Crashed.

Good catch. I'll see if I can fix that too.

Carlos

---
* Origin: cyberiada-NNTP (2:341/234.99)