Re: [S-mailx] S-nail segfaults and dumps a core
MKCITRUS=no was set, but I don't think it's usable without a Makefile patch, so it's built-in. Check out this bit from the libc Makefile[0]: -- workaround for I18N stuffs: build singlebyte setlocale() for libc.a, # multibyte for libc.so. the quirk should be removed when we support # dlopen() from within statically linked binaries. CSHLIBFLAGS+= -D_I18N_DYNAMIC" -- Could be related to other locale problems I mentioned in a current-users@ thread[1]. So, I guess it's not related to s-nail. [0]: https://github.com/NetBSD/src/blob/trunk/lib/libc/Makefile#L154 [1]: http://mail-index.netbsd.org/current-users/2018/11/05/msg034545.html -- caóc
Re: [S-mailx] S-nail segfaults and dumps a core
Good evening Cág, Cág wrote in <0d1d85a14893b88d0c892c0b4a36f...@bitmessage.ch>: |> Please excuse the late reply. | |Well, not so late. oh, yes it is. |> I did not know this thread was unfinished. |> But i am also entirely out of time, and fallen there too. | |Back then you asked about the system configuration, and so I didn't |want this to be unanswered. I .. see. :) |>> ERROR: MULTIBYTE: 0/n/false/no/off or 1/y/true/yes/on/require, got: |>> *** Error code 1 |> I do not think that this error message comes from S-nail? | |The question can be asked another way: |What check is it doing at that stage? What does it check for? The answer, my friend, is blowing in the wind! |> NetBSD multibyte via citrus and static, does this work at all?? |> My own one, well... | |I might be wrong, but I think I set CITRUS=no in mk.conf... |Gonna check it tomorrow. | |> And i am sorry, i cannot do any testing in the following weeks, |> i have two new (used) laptops which now drive AlpineLinux on bare |> metal -- the first Linux on bare metal since 2002 i guess! | |Were you running Linux only on VMs or what? Ja., since about 2002 only FreeBSD and since about 2009 Mac OS on bare metal. There is also OpenBSD on my old Athlon Box if i recall correctly, but i have not booted it in a long time. |Bis bald It will be my pleasure. Ciao. --steffen | |Der Kragenbaer,The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
Re: [S-mailx] S-nail segfaults and dumps a core
Steffen Nurpmeso wrote: Hello Dohle! Please excuse the late reply. Well, not so late. I did not know this thread was unfinished. But i am also entirely out of time, and fallen there too. Back then you asked about the system configuration, and so I didn't want this to be unanswered. ERROR: MULTIBYTE: 0/n/false/no/off or 1/y/true/yes/on/require, got: *** Error code 1 I do not think that this error message comes from S-nail? The question can be asked another way: What check is it doing at that stage? What does it check for? NetBSD multibyte via citrus and static, does this work at all?? My own one, well... I might be wrong, but I think I set CITRUS=no in mk.conf... Gonna check it tomorrow. And i am sorry, i cannot do any testing in the following weeks, i have two new (used) laptops which now drive AlpineLinux on bare metal -- the first Linux on bare metal since 2002 i guess! Were you running Linux only on VMs or what? Bis bald -- caóc
Re: [S-mailx] S-nail segfaults and dumps a core
Hello Dohle! Please excuse the late reply. Cág wrote in : |Sorry for such a long delay -- really busy days. I did not know this thread was unfinished. But i am also entirely out of time, and fallen there too. |I'm now using the latest -CURRENT, and trying to build s-nail |fails at this part: | |-- |ERROR: MULTIBYTE: 0/n/false/no/off or 1/y/true/yes/on/require, got: |*** Error code 1 |-- | |The system is NetBSD 8.99.25, everything is linked statically. |The compiler is Clang. I noticed several problems, and I guess they | |are due to static linking. On the other hand they could be due to me |making a mistake somewhere. | |Does the error above relate to locales or is it a curses error? I do not think that this error message comes from S-nail? Static linking, puuhh, that is a hard one nowadays! NetBSD multibyte via citrus and static, does this work at all?? My own one, well... And i am sorry, i cannot do any testing in the following weeks, i have two new (used) laptops which now drive AlpineLinux on bare metal -- the first Linux on bare metal since 2002 i guess!, and there are so many unsolved problems (WLAN at 128 KB/s, usb-storage quirks not doing, upgrade to BTRFS still not done, need to create Xmodmap for english keyboard which gives me Unicode combining, for at least some, but better all). It is plain terrible! I hope i will be able to have some VMs back up in December! I need to take some time for development, to have some positive progress somewhere, at least. ^_^ Ciao! --steffen | |Der Kragenbaer,The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
Re: [S-mailx] S-nail segfaults and dumps a core
Hi Steffen, Sorry for such a long delay -- really busy days. I'm now using the latest -CURRENT, and trying to build s-nail fails at this part: -- ERROR: MULTIBYTE: 0/n/false/no/off or 1/y/true/yes/on/require, got: *** Error code 1 -- The system is NetBSD 8.99.25, everything is linked statically. The compiler is Clang. I noticed several problems, and I guess they are due to static linking. On the other hand they could be due to me making a mistake somewhere. Does the error above relate to locales or is it a curses error? Thanks! -- caóc
Re: [S-mailx] S-nail segfaults and dumps a core
Hi Dohle. Cág wrote in <20180930214246.3s2mj%c...@bitmessage.ch>: |I'll report as soon as I can. As for now, don't worry! What was the problem, actually? For me it will be hard to do anything real about that, until we come to v15: not before 2020. Maybe i can add some expensive hacks around the stuff, and partially use SysV signal handling at this specific place, but in general not, because the codebase uses OpenSSL on the sockets directly, instead of doing the network I/O itself and using OpenSSL via memory-based I/O: v15. Sorry for that. (I still could not cause a crash.) --steffen | |Der Kragenbaer,The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
Re: [S-mailx] S-nail segfaults and dumps a core
Steffen, I'll report as soon as I can. As for now, don't worry! -- caóc
Re: [S-mailx] S-nail segfaults and dumps a core
Steffen Nurpmeso wrote in <20180927232839.guggp%stef...@sdaoden.eu>: |Cág wrote in <20180927225715.iv3ds%c...@bitmessage.ch>: ||Steffen Nurpmeso wrote: ||> I cannot reproduce it at a glance via AlpineLinux: ||> [...] ||I forgot to mention that I use s-nail with torify, but the same seems | |You happen to use tor via *socks-proxy*? That is pretty cool, ... ||Interrupting this operation may turn the DNS resolver unusable ||^Cs-nail interrupted Actually, if you use *socks-proxy*, we only look up the host name of the proxy iirc, the real connection end point will be done by the SOCKS proxy, so this can be avoided by adding an entry to /etc/hosts, or by using an IP address, which the C library resolver should treat via the logical short path. |A nice weekend, Dohle! Ciao! --steffen | |Der Kragenbaer,The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
Re: [S-mailx] S-nail segfaults and dumps a core
Hallo. Cág wrote in <20180927225715.iv3ds%c...@bitmessage.ch>: |Steffen Nurpmeso wrote: |> I cannot reproduce it at a glance via AlpineLinux: |> [...] |> It seems i have to boot my NetBSD 8 VM, it could be that the fix |> takes until tomorrow, thus. (Very slow, dependent on how weird it |> is...) | |I forgot to mention that I use s-nail with torify, but the same seems You happen to use tor via *socks-proxy*? That is pretty cool, i still have not tried to use tor at all, even though i am tracking the announcements and the source code (since "Import v0.3.0.7, 2017-05-26", but the heck!, just cannot find any time). I have tested *socks-proxy* only via ssh(1)... |to happen without it. Also it's not the same all the time: I reconnected |with `dhclient -r`, then the usual wpa_supplicant+dhclient thing, s-nail |just hanged after "G": Yeah, as i wrote in the other mail, set *verbose* to see the progress ticks. I will change that and activate them in interactive mode automatically. |-- |^C |Interrupting this operation may turn the DNS resolver unusable |^Cs-nail interrupted Uh!!! This is hard :). |Interrupt |There are new messages in the error [...] |[...] |ERROR# ? q |Held messages in [my inbox] |[1] Segmentation fault (core dumped) s-nail I am very sorry Cág, but until v15 this codebase uses siglongjmp(), and except for libjp(e???)g each and every library i know will occasionally just break after such a longjmp. There is nothing that can be done here, but it is all my fault, i should have set back in 2012 and refrain from working with the nail codebase as such, instead i should have taken the OpenBSD Mail and develop from there on, and all this signal mess would simply not exist. This is v15, with SysV signal handling. Until then the above can happen. This is actually stated in the manual sections CAVEATS. |Maybe make flags should be noted [shameless stolen from the APKBUILD]: |VAL_IDNA=idn |VAL_RANDOM="libgetrandom sysgetrandom urandom builtin" Hmm. Likely better to not use such things at all unless you want a fixed configuration; NetBSD has arc4random: . VAL_RANDOM: arc4random(3) ... yes but forcing usage of the system OpenSSL may be an option VAL_RANDOM=tls,arc4 Not to talk about the unscientific builtin arc4, but well, needless, needless... |There are patches also used, I am attaching them, though they don't seem |to be related to the issue, and they don't exist in aports. Should I |remove them? Yes! Yes!! Yes!!! How could you apply them to v14.9.11 I would nonetheless be interested in the NYD, i think i will try and see what i can do to make this better on NetBSD. That we crash in non-network code after such an interruption, that is very very bad! |Thanks, Steffen! A nice weekend, Dohle! --steffen | |Der Kragenbaer,The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
Re: [S-mailx] S-nail segfaults and dumps a core
Hello again Cág. Steffen Nurpmeso wrote in <20180927223606.qal28%stef...@sdaoden.eu>: |Steffen Nurpmeso wrote in <20180927222911.jvncp%stef...@sdaoden.eu>: ||Cág wrote in <20180927215317.sicaq%c...@bitmessage.ch>: |||Suppose you don't have an Internet connection. Then you try something |||like this "file pop3s://user%40mailprovider etc etc" (I have Should be pop3s://user@mailprovider btw. I agree that `urlcodec' should be much smarter, and not do #?0!0/NONE[#+[..]? urlenc pop3s://user@mailprovider in: (25 bytes) out: (33 bytes) That is terrible :-( Ach! one more problem which will linger until v15 when URL is a real object and you can say url_open_connection(urlobject) or such, like url_conn_factory() or whatever. :-( |||"commandalias G" for that). In this case s-nail produces a segmentation |||fault and dumps a core that backtraces to libc. Why does it happen and |||should it? ... |||My system, if this matters, NetBSD. || ||I hope not. | |I cannot reproduce it at a glance via AlpineLinux: And i also cannot reproduce it on NetBSD 8. ..However, instead of failing immediately with NETUNREACH NetBSD tries to connect (from within the VM), which looks like as if it would be hanging. I wonder whether the connection progress which will appear when *verbose* is set shall always appear in interactive mode. Hmm. I think i will do this. .. #?0!0/NONE[#/var/mail/steffen]? G s-nail: Resolving host :pop3s ... done s-nail: Connecting to ...:pop3s ... timeout s-nail: Could not connect: Operation timed out ERROR# #?1!60/TIMEDOUT[#/var/mail/steffen]? I wonder Cág, which NetBSD version, which compiler... What does the follwing command say, for example: $ s-nail -Sverbose -Xversion -Xx Could you compile with debug enabled and send the generated NYD output when the program crashes (compress the file, my postfix has a size limit and it compresses very good anyway), or could you give me the "bt" backtrace that gdb produces, please? This would be very helpful! Segfault, and that on a Friday. Winter time about to come it seems, grrr, terrible!!! --steffen | |Der Kragenbaer,The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
Re: [S-mailx] S-nail segfaults and dumps a core
Steffen Nurpmeso wrote in <20180927222911.jvncp%stef...@sdaoden.eu>: |Cág wrote in <20180927215317.sicaq%c...@bitmessage.ch>: ||Suppose you don't have an Internet connection. Then you try something ||like this "file pop3s://user%40mailprovider etc etc" (I have ||"commandalias G" for that). In this case s-nail produces a segmentation ||fault and dumps a core that backtraces to libc. Why does it happen and ||should it? | |I will investigate, and _definetely_ not! |I am unpleasantly surprised, i have tested that in the past for |sure. | ||My system, if this matters, NetBSD. | |I hope not. I cannot reproduce it at a glance via AlpineLinux: #?0!0/NONE[st#/var/spool/mail/steffen]? commandalias G file pop #?0!0/NONE[st#/var/spool/mail/steffen]? G You need a passphrase to unlock the secret key for user: "Steffen Nurpmeso " 4096-bit RSA key, ID A57802DD, created 2017-11-30 (main key ID 1883A0DD) mail: Could not connect: Network is unreachable There are new messages in the error message ring (denoted by ERROR) The `errors' command manages this message ring ERROR# #?1!101/NETUNREACH[st#/var/spool/mail/steffen]? It seems i have to boot my NetBSD 8 VM, it could be that the fix takes until tomorrow, thus. (Very slow, dependent on how weird it is...) ||Thanks! All in return! --steffen | |Der Kragenbaer,The moon bear, |der holt sich munter he cheerfully and one by one |einen nach dem anderen runter wa.ks himself off |(By Robert Gernhardt)
