@Fab & kas1e

Quote:

Anyway, my guess if you have an problem with the way SocketBase is handled in threads or so (or some reentrance issue with Roadshow GetHostByName, maybe...

Yes, on OS4 SocketBase (or it's Iface) cannot be shared between tasks. It has to be opened for each task that uses it. I just remembered that.

Or at least you need to call some special function to make it thread-safe or sharable, or to share a socket. It can be done but it isn't automatic. That is probably why we never got that threaded curl to work...

_________________
- Don't get fooled by my avatar, I'm not like that (anymore, mostly... maybe only sometimes)
> Amiga Classic and OS4 developer for OnyxSoft.

@kas1e

Do you have any planned release date for OWB 1.17 in your mind ?
Like June, July ?

_________________
AmigaOS4 Beta-Tester
----------------------------------
Amiga X5000 8GB Radeon RX 560
Amiga A1222 2GB Radeon RX 550
Sam 460 2GB RadeonRX 550

@Deniil715

Quote:

Yes, on OS4 SocketBase (or it's Iface) cannot be shared between tasks. It has to be opened for each task that uses it. I just remembered that. Or at least you need to call some special function to make it thread-safe or sharable, or to share a socket. It can be done but it isn't automatic. That is probably why we never got that threaded curl to work...

Sounds easy, we can just open/close socket base on starting/ending of tasks, and hope it will fix everything. Anyway, i already start to annoy Olaf about some help..

SinanG
Quote:

Do you have any planned release date for OWB 1.17 in your mind ? Like June, July ?

Plan is to make threaded curl + owb1.9 work fine firstly. Then we can move on. Because blocking of GUI suck, and while we have source code already, do not port/resuse it normally suck in twice.

So, once it will be done, then in can be used with 1.17 and then add all the stuff like custom mouse pointers, mouse wheel, custom tabs and can be released for tests.

When - dunno. Just once we will fix problems with curl, i think all will be close.

Last edited by kas1e on 29-May-2012 at 02:24 PM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

I'm not sure if it's the same but latest Reaction version (3.32) add a threaded resolve within LibCURL, i didn't test so mutch because the final result is that the GUI will be stalled for many seconds each time you will close a single tab or the program .. but atleast it never crash

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@samo79

Quote:

I'm not sure if it's the same but latest Reaction version (3.32) add a threaded resolve within LibCURL, i didn't test so mutch because the final result is that the GUI will be stalled for many seconds each time you will close a single tab or the program .. but atleast it never crash

What you mean by "the same" ? If it didn't crash then its not the same for sure. And in general, does not matter what ra-owb have , because its all close-source, and we can't check sources of it even for interest. + If it buggy, its make even less sence.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

I meant if the threaded resolve introduced in OWB 3.32 was added exactly to prevent the blocking of the GUI in the event there are connection problems or just if it was added for somethings else ... to be honest i don't understand why it was added as i can't see any difference in general if a connection is down for any reason or not, more now it will stall the GUI also when the connection is ok, for example as i say when you close a tab or the program (but it doesn't crash aniway)

Quote:

If it didn't crash then its not the same for sure

Yep but if it was added for the same reason (to prevent the GUI blocking) atleast we can be sure that it's possible to add it without any crash from bsdsocket and so on

Quote:

And in general, does not matter what ra-owb have , because its all close-source, and we can't check sources of it even for interest. + If it buggy, its make even less sence.

Yep .. :-/

Last edited by samo79 on 29-May-2012 at 04:33 PM.

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@Deniil715

Actually, it's a bit wrong: on OS4, roadshow allows to share sockets across threads and socketbase as well (which is the only reason why something like Wesnoth has network working). It might not be default behaviour, though, i don't know.

On AmigaOS3.x, MorphOS and AROS, sockets can't be shared across threads (be it with MIAMI, Netstack, AmiTCP; ...) (ObtainSocket/ReleaseCopyOfSocket are required to do so), and SocketBase must be opened for each thread. This is of course what i do in my CURL port, anyway, so it shouldn't be a problem, unless there's some additional magic required regarding interfaces stuff.

Last edited by Fab on 29-May-2012 at 04:43 PM.

@kas1e / samo79

OWB 3.32 with threaded libcurl does crash - there were loads of reports of it at the time it was released. Curiously some people never had problems.

It would be nice to get a working threaded libcurl, and if changes need to be made in libcurl itself (which is looking increasingly unlikely from what has been said), then get those fed back into the project too.

edit Just had a quick look at the bsdsocket autodocs.
SocketBaseTagList() SBTC_CAN_SHARE_LIBRARY_BASES looks very interesting!

Last edited by Chris_Y on 29-May-2012 at 06:55 PM.

_________________
"Miracles we do at once, the impossible takes a little longer" - AJS on Hyperion
Avatar is Tabitha by Eric W Schwartz

@Chris_Y (and others)

For testing purposes i build owb1.9 with pthreaded curl. I.e. that curl which have no changes from fab, just original code, which just builds with threaded resolver. And new binary linked against it : there is new, pthreaded binary.

As for me, its reacts just the same as one from fab: on begining i have no crashes (not like for most of you, only for Dragster its like for me), and then, after 3-4 blocking connection i have 1:1 the same crash in Curl_ipv4_resolve_r() function like a have from fab's curl.

Check this out, interesting how it will reacts and will it or not crashes on first visited url. Because if not, then its our port-code problems for that crash. While second crash its something os4 specific.

Last edited by kas1e on 29-May-2012 at 09:40 PM.
Last edited by kas1e on 29-May-2012 at 09:35 PM.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

It works so far, posting from it...

_________________
Sam440ep-flex 800MHz 1GB RAM R9250 128MB SB Live!

@kas1e

After many ( +-20) urls I got this...

Crash log for task "OWB"
Generated by GrimReaper 53.2
Crash occured in module bsdsocket.library at address 0x6FC02EB8
Type of crash: DSI (Data Storage Interrupt) exception

Register dump:
GPR (General Purpose Registers):
0: 00000000 50193930 00000000 00000000 00000000 00000000 021AA360 021AA3D4
8: 00000001 00000000 00000001 00000000 22482024 00000000 501939AC 50193A18
16: 00000000 00000400 02270000 51263464 591B20EE 51FA0672 FFFFFFFF 00000000
24: 00000000 00000000 501939B4 00000000 00000400 00000000 51FA0644 00000000


FPR (Floating Point Registers, NaN = Not a Number):
0: nan -6.31401e-213 -3.94623e-214 1.74578e+306
4: -1.54147e-216 1.50466e-36 -9.63414e-218 -2.37995e-292
8: 2.00706e+101 -1.06907e-211 -9.86554e-215 -5.45273e-290
12: 1.38039e+70 -3.40793e-291 0 4.34326e+13
16: 4.90817e-193 0 0 0
20: 0 0 0 0
24: 0 2.65249e-315 0 0
28: 0 0 0 5.58058e-222

FPSCR (Floating Point Status and Control Register): 0x82000000


SPRs (Special Purpose Registers):
Machine State (msr) : 0x0002F030
Condition (cr) : 0x22482022
Instruction Pointer (ip) : 0x6FC02EB8
Xtended Exception (xer) : 0x00000000
Count (ctr) : 0x01812660
Link (lr) : 0x6FC02F0C
DSI Status (dsisr) : 0x00000000
Data Address (dar) : 0x00000008



680x0 emulated registers:
DATA: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
ADDR: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
FPU0: 0 0 0 0
FPU4: 0 0 0 0



Symbol info:
Instruction pointer 0x6FC02EB8 belongs to module "bsdsocket.library" (HUNK/Kickstart)

Stack trace:
module LIBS:bsdsocket.library at 0x6FC02EB8 (section 5 @ 0x4CE98)
module LIBS:bsdsocket.library at 0x6FC051E8 (section 5 @ 0x4F1C8)
module LIBS:bsdsocket.library at 0x6FC055B0 (section 5 @ 0x4F590)
module LIBS:bsdsocket.library at 0x6FBE7FBC (section 5 @ 0x31F9C)
module LIBS:bsdsocket.library at 0x6FBF1A04 (section 5 @ 0x3B9E4)
module LIBS:bsdsocket.library at 0x6FBF0A88 (section 5 @ 0x3AA68)
module LIBS:bsdsocket.library at 0x6FBF0CE4 (section 5 @ 0x3ACC4)
module LIBS:bsdsocket.library at 0x6FBF125C (section 5 @ 0x3B23C)
module LIBS:bsdsocket.library at 0x6FBE36E8 (section 5 @ 0x2D6C8)
native kernel module newlib.library.kmod+0x00039970
OWB:Curl_ipv4_resolve_r()+0x7C (section 1 @ 0xe0f88c)
OWB:gethostbyname_thread()+0x20 (section 1 @ 0xdf1a38)
OWB:curl_thread_create_thunk()+0x38 (section 1 @ 0xe17c4c)
pthreads.library:run()+0x278 (section 1 @ 0x4134)
pthreads.library:ThreadCode()+0x35C (section 1 @ 0x44dc)
native kernel module dos.library.kmod+0x00022a0c
native kernel module kernel+0x0003af48
native kernel module kernel+0x0003afc8


PPC disassembly:
6fc02eb0: 40820014 bne- 0x6FC02EC4
6fc02eb4: 813e000c lwz r9,12(r30)
*6fc02eb8: a0090008 lhz r0,8(r9)
6fc02ebc: 70090004 andi. r9,r0,4
6fc02ec0: 40820190 bne- 0x6FC03050

System information:

CPU
Model: AMCC PPC440EP V1.3
CPU speed: 799 MHz
FSB speed: 133 MHz
Extensions:

Machine
Machine name: Sam440EP
Memory: 1048576 KB
Extensions: bus.pci

_________________
Sam440ep-flex 800MHz 1GB RAM R9250 128MB SB Live!

@mr2

So seems crash on first connection its our-port-related, while second , in Curl_ipv4_resolve_r() are same for everything ..

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

I will test this binary more intensively tonight... I'll report back.


Great work, btw

Cheers,

Dragster

_________________
Powerbook G4 1139/MOS 3.9/OSX 10.5.8. Pegasos II 2B5 Mobo. 2 GB RAM. 400 GB x 2 GB ATA HD, USB 2.0 PCI. SoundBlaster Audigy 2ZS Gold. Quad boot: AmigaOS 4.1 FInal Edition, MorphOS 3.9 regged, Debian Lenny PPC, OpenSUSE 11.1. Rockin'

@kas1e

This new binary works on my Sam460 + OS4.1.5 Beta...I visited more than 20 urls at the moment..still no crash..

But this version stalls in some web pages (images took a lot time to load also that slows down the other tabs of OWB.)

_________________
AmigaOS4 Beta-Tester
----------------------------------
Amiga X5000 8GB Radeon RX 560
Amiga A1222 2GB Radeon RX 550
Sam 460 2GB RadeonRX 550

@kas1e

Quote:

But this version stalls in some web pages (images took a lot time to load also that slows down the other tabs of OWB.)

Yes, the same here.

_________________
Sam440ep-flex 800MHz 1GB RAM R9250 128MB SB Live!

EDIT1: with the first owb1.9 version attahced, now trying with the 2nd version you attahced kasle.

Hi too get a GR (DSI) after type some URL (or in search web string gadget)

crashlog:
Crash log for task "owb"
Generated by GrimReaper 53.5
Crash occured in module bsdsocket.library at address 0x7FD3BCC8
Type of crash: DSI (Data Storage Interrupt) exception

Register dump:
GPR (General Purpose Registers):
0: 00000003 6A0C5180 00000000 6A0C518C 00000001 00000004 00000001 6FF91000
8: 6A0C5398 6A0C518C 7FCF69B0 6A0C518C 6A0C5230 6A26B3D8 00000000 00000000
16: 00000000 7E957240 00000000 00000002 00000000 6814F508 6A0C5768 6A0C5708
24: 00000000 6A0C518C 6B37B4E8 6A0C524C 00000019 00000004 6AD757CC 8004667E


FPR (Floating Point Registers, NaN = Not a Number):
0: nan 0 0 -0
4: -0 0 1 0
8: 0 -0.00196464 3.8598e-06 292
12: 4.5036e+15 0 0 0
16: 0 0 0 0
20: 0 0 0 0
24: 0 0 0 0
28: 0 0 1.33838e+09 1.33838e+09

FPSCR (Floating Point Status and Control Register): 0x82002000


SPRs (Special Purpose Registers):
Machine State (msr) : 0x0200F030
Condition (cr) : 0x44822028
Instruction Pointer (ip) : 0x7FD3BCC8
Xtended Exception (xer) : 0x00000000
Count (ctr) : 0x00000004
Link (lr) : 0x7FD22D70
DSI Status (dsisr) : 0x40000000
Data Address (dar) : 0x00000001



680x0 emulated registers:
DATA: 6C0AB1FE 00000003 00000000 00000000 00000000 00000000 00000000 00000000
ADDR: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 6A0C5370
FPU0: 0 0 0 0
FPU4: 0 0 0 0



Symbol info:
Instruction pointer 0x7FD3BCC8 belongs to module "bsdsocket.library" (HUNK/Kickstart)

Stack trace:
module LIBS:bsdsocket.library at 0x7FD3BCC8 (section 5 @ 0x53CAC)
module LIBS:bsdsocket.library at 0x7FD114F0 (section 5 @ 0x294D4)
owb:curlx_nonblock()+0x38 (section 1 @ 0xDF1C60)
owb:singleipconnect()+0x30c (section 1 @ 0xDE9BAC)
owb:Curl_connecthost()+0x138 (section 1 @ 0xDEA15C)
owb:Curl_setup_conn()+0x260 (section 1 @ 0xDFCC68)
owb:Curl_async_resolved()+0x38 (section 1 @ 0xDEEEC4)
owb:multi_runsingle()+0x79c (section 1 @ 0xDECBA0)
owb:curl_multi_perform()+0x120 (section 1 @ 0xDED68C)
owb:_ZN7WebCore21ResourceHandleManager21downloadTimerCallbackEPNS_5TimerIS0_EE()+0x170 (section 1 @ 0xFD16C)
owb:_ZN7WebCore5TimerINS_21ResourceHandleManagerEE5firedEv()+0x30 (section 1 @ 0xFE584)
owb:_ZN7WebCore12ThreadTimers24sharedTimerFiredInternalEv()+0x140 (section 1 @ 0x11A910)
owb:_ZN7WebCore12ThreadTimers16sharedTimerFiredEv()+0x3c (section 1 @ 0x11A974)
owb:_ZN7WebCore17fireTimerIfNeededEv()+0x30 (section 1 @ 0x10F654)
owb:_ZN14WebViewPrivate21fireWebKitTimerEventsEv()+0x10 (section 1 @ 0xA5284)
owb:_ZN7WebView21fireWebKitTimerEventsEv()+0x14 (section 1 @ 0x88260)
owb:_ZL28handleMM_OWBApp_WebKitEventsP6IClassPmP4_Msg()+0x78 (section 1 @ 0x4AC8)
owb:_ZL8dispatchP6IClassPmP4_Msg()+0x310 (section 1 @ 0xC8C4)
muimaster.library:CustomClassDispatcher()+0xa0 (section 1 @ 0x40F8)
native kernel module intuition.library.kmod+0x00018178
native kernel module intuition.library.kmod+0x0001839c
native kernel module intuition.library.kmod+0x00008424
native kernel module intuition.library.kmod+0x00008064
owb:_Z9main_loopv()+0x228 (section 1 @ 0x11D8)
owb:main()+0x12c (section 1 @ 0x134C)
native kernel module newlib.library.kmod+0x0000208c
native kernel module newlib.library.kmod+0x00002d34
native kernel module newlib.library.kmod+0x00002eb0
owb:_start()+0x170 (section 1 @ 0x170)
native kernel module dos.library.kmod+0x00022838
native kernel module kernel+0x000416f0
native kernel module kernel+0x00041770


PPC disassembly:
7fd3bcc0: 7ca903a6 mtctr r5
7fd3bcc4: 7d695b78 mr r9,r11
*7fd3bcc8: 88040000 lbz r0,0(r4)
7fd3bccc: 38840001 addi r4,r4,1
7fd3bcd0: 98090000 stb r0,0(r9)

System information:

CPU
Model: Motorola MPC 7445/7455 Apollo V2.1
CPU speed: 800 MHz
FSB speed: 133 MHz
Extensions: performancemonitor altivec

Machine
Machine name: AmigaOne
Memory: 524288 KB
Extensions: bus.pci bus.agp

Tasks
..
TCP/IP Control (Waiting)
Stack: 0x6b312004 - 0x6b321ffc, pointer @ 0x6b321d70 (Cookie OK)
Signals: SigRec 0x00000010, SigWait 0x40000000
State: Process (Waiting)
[OWB] IconDatabase (Waiting)
Stack: 0x69ea4004 - 0x69f98ffc, pointer @ 0x69f98eb0 (Cookie OK)
Signals: SigRec 0x00000010, SigWait 0x00000000
State: Process (Waiting)
TCP/IP Superserver [TCP/IP Superserver] (Waiting)
Stack: 0x6b32e004 - 0x6b33dffc, pointer @ 0x6b33d9f0 (Cookie OK)
Signals: SigRec 0xd0000080, SigWait 0x00000000
State: Process (Waiting)
TCP/IP Configuration (Waiting)
Stack: 0x6b34a004 - 0x6b359ffc, pointer @ 0x6b359e10 (Cookie OK)
Signals: SigRec 0xf8003000, SigWait 0x00000000
State: Process (Waiting)
..
TCP/IP Log (Waiting)
Stack: 0x6b366004 - 0x6b375ffc, pointer @ 0x6b375f00 (Cookie OK)
Signals: SigRec 0x80003000, SigWait 0x00000000
State: Process (Waiting)
..
owb (Crashed)
Stack: 0x69fd1004 - 0x6a0c5ffc, pointer @ 0x6a0c5180 (Cookie OK)
Signals: SigRec 0x0f80d000, SigWait 0x0c200100
State: Process (Crashed)


Hope it helps

Last edited by jabirulo on 30-May-2012 at 02:05 PM.

@jabirulo

replying to myself

Tested the 2nd OWB1.9 attached and now it doesn't crash on first visited URL.
Will test it further to see when I get GR

TIA

@Fab

Quote:

Actually, it's a bit wrong: on OS4, roadshow allows to share sockets across threads and socketbase as well (which is the only reason why something like Wesnoth has network working). It might not be default behaviour, though, i don't know.

Not default, no. Some magic is needed I think, like reopening the interfaces and flag a socket as sharable or something. Haven't looked at in in years since I forgot about it.

@Chris_Y

Quote:

SocketBaseTagList() SBTC_CAN_SHARE_LIBRARY_BASES looks very interesting!

That might be what we need.

_________________
- Don't get fooled by my avatar, I'm not like that (anymore, mostly... maybe only sometimes)
> Amiga Classic and OS4 developer for OnyxSoft.

@Deniil715

Well, ok. But my code doesn't rely on sharing bases, though. It opens the library in each spawned dns resolution process. There's also no shared sockets there.

@kas1e

I hope that you can solve all that threaded curk problems soon, however OWB 1.17 waqs so faster than 1.9 than I thinbk that if was possible to have a non threaded version was however better than current 1.9 that ahve also issues with Google Images/youtube and so on...

Naturally if the work on 1.17 not threaded was easy and fast...

_________________
Simone"Tuxedo"Monsignori, Perugia, ITALY.