Pike-devel

pike-devel@lists.lysator.liu.se

1 participants
3041 discussions

Need assistance decoding a gc fatal
by H William Welliver 13 Feb '21

13 Feb '21

I’ve been working on updating/enhancing the debugger concept code and I’ve run into a situation where a gc fatal report is generated. I’m not exactly sure what it’s trying to say, as I /think/ the data all makes sense (though I’m unclear what 2+0 internal refs vs 2 refs might mean. The pike in question is based on master from quite a while ago. I tried merging current master (which itself was a surprisingly easy task), but pike fails to compile due to typechecker errors. I tried building straight from master, but the build similarly fails with --with-debug and without that, pike compiles but fails to start. I’m hoping that doesn’t prevent coming up with an explanation for the problem. I believe the “thing” in question is the main program, which comes from a file called file.pike. I’ve created a breakpoint object( for a line in this program), which is a PIKECLASS defined in _Debug/debug.cmod. The breakpoint object /does/ hold a reference to the program containing the breakpoint, but at this point, the breakpoint object hasn’t been destroyed. Can anyone detect a problem from this report? Bill ---CUT HERE --- /Users/hww3/pikedbg/src/gc.c:1938: GC fatal: Thing is getting more internal refs (2 + 0) than refs (2). (Could be an extra free somewhere, or a pointer might have been checked more than once.) **Block: 0x10dd303f0 Type: program Refs: 2 **Got gc marker at 0x10dd303f0: flags=0x00002 refs=2 weak=0 xrefs=0 saved=2 frame=0x0 **Program id: 65669, flags: f, parent id: -1 **Location: /Users/hww3/pikedbg/file.pike:0 **Identifiers: ** 0 fun: name: main addr: 0x24 ** 1 fun: use name: f addr: 0x360 ******************* **Looking for references to 0x10dd303f0: ***One ref found as mapping index. -> from mapping 0x7faf328004d0 offset -8362904 **In index program(/Users/hww3/pikedbg/file.pike) **Block: 0x7faf328004d0 Type: mapping Refs: 1 Refs=1, next=0x7faf32800490, prev=0x7faf32800510, flags=0x0, size=33, hashsize=16 Indices type field = program Values type field = string ([ /* 33 elements */ master: "/master", object_program(Getopt): "/Users/hww3/pikedbg/lib/modules/Getopt.pmod", /main: "/main", object_program(Debug): "/Users/hww3/pikedbg/lib/modules/Debug.pmod/module.pmod", object_program(Debug.Debugger): "/Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/module.pmod", Debug.Debugger.Breakpoint: "/Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/Breakpoint.pike", Debug.Debugger.DebugServer: "/Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/DebugServer.pike", object_program(Debug.Debugger.DebugAdapterProtocol): "/Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/DebugAdapterProtocol.pmod", - - - trimmed for brevity ]) ******************* ----------end------------ ***One ref found as mapping value. -> from mapping 0x7faf328003d0 offset 537720 **In value with index "/main" **Block: 0x7faf328003d0 Type: mapping Refs: 1 Refs=1, next=0x7faf32800290, prev=0x7faf32800350, flags=0x0, size=33, hashsize=16 Indices type field = string Values type field = program ([ /* 33 elements */ "/Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/Breakpoint.pike": Debug.Debugger.Breakpoint, "/Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/DebugAdapterProtocol.pmod": object_program(Debug.Debugger.DebugAdapterProtocol), "/Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/DebugServer.pike": Debug.Debugger.DebugServer, "/Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/module.pmod": object_program(Debug.Debugger), "/Users/hww3/pikedbg/lib/modules/Debug.pmod/module.pmod": object_program(Debug), "/Users/hww3/pikedbg/lib/modules/Pike.pmod/module.pmod": object_program(Pike), - - - trimmed for brevity "/main": /main, "/master": master ]) ******************* ----------end------------ ***One ref found. -> from object 0x7faf3286eaa0 offset 10033200 **In storage for inherit 0 **In variable prog **In storage for inherit 1 (debug_breakpoint) **Block: 0x7faf3286eaa0 Type: object Refs: 1 **Got gc marker at 0x7faf3286eaa0: flags=0x00002 refs=1 weak=0 xrefs=0 saved=1 frame=0x0 **Program id: 65699 **Object variables: ** === In inherit "debug_breakpoint", program 65626: ** rtt: program name: prog off: 0 value: program(/Users/hww3/pikedbg/file.pike) ** rtt: string name: programs_fname off: 8 value: "/Users/hww3/pikedbg/file.pike" ** rtt: string name: real_path off: 16 value: "/Users/hww3/pikedbg/file.pike" ** rtt: string name: within_fname off: 24 value: "/Users/hww3/pikedbg/file.pike" ** rtt: int name: line_number off: 32 value: 18 ** rtt: int name: enabled off: 40 value: 1 ** rtt: int name: want_enabled off: 48 value: 0 **Describing program 0x10dd33000 of object: **Got gc marker at 0x10dd33000: flags=0x00002 refs=4 weak=0 xrefs=0 saved=4 frame=0x0 **Program id: 65699, flags: 202f, parent id: -1 **The program was written in C. **Location: /Users/hww3/pikedbg/lib/modules/Debug.pmod/Debugger.pmod/Breakpoint.pike:0 **There is no parent (any longer?) ******************* ----------end------------ ***One ref found as the program of an object. **Block: 0x7faf3286ea50 Type: object Refs: 5 **Program id: 65669 **Object variables: ** (none) **Describing program 0x10dd303f0 of object: **Got gc marker at 0x10dd303f0: flags=0x00002 refs=2 weak=0 xrefs=0 saved=2 frame=0x0 **Program id: 65669, flags: f, parent id: -1 **Location: /Users/hww3/pikedbg/file.pike:0 **There is no parent (any longer?) ******************* ----------end------------ ***One ref found in current_program in frame on stack. **Block: 0x7faf32821b38 Type: pike_frame Refs: 1 **Function f at /Users/hww3/pikedbg/file.pike:14 ******************* ----------end------------ ***One ref found in current_program in frame on stack. **Block: 0x7faf32821aa0 Type: pike_frame Refs: 1 **Function main at /Users/hww3/pikedbg/file.pike:2 ******************* ----------end------------ ***One ref found on svalue stack. -> from pike_frame 0x7faf32821a08 offset -140385920272696 **At position 12 among locals **Block: 0x7faf32821a08 Type: pike_frame Refs: 1 **Function _main at /Users/hww3/pikedbg/build/darwin-18.7.0-x86_64/master.pike:4069 ******************* ----------end------------ **Done looking for references to 0x10dd303f0, found 7 refs. Pike was in GC stage 100 when this fatal occurred. Backtrace at time of fatal: src/modules/_Debug/debug.cmod:652: lib/modules/Debug.pmod/Debugger.pmod/Breakpoint.pike() -:1: Stdio.FILE(2, "w", 777 /* fd=2 */)->write("\nSending response: \nContent-Length: 228\r\n\r\n{\"request_seq\":16,\"seq\":0,\"body\":{\"threads\":[{\"name\":\"Thread.Thread(4541273536)\",\"id\":4541273536},{\"nam"+[80]+"\"message\":null,\"success\":1,\"type\":\"response\"}\n") lib/modules/Stdio.pmod/module.pmod:1105: Stdio.FILE(2, "w", 777 /* fd=2 */)->write("\nSending response: \nContent-Length: 228\r\n\r\n{\"request_seq\":16,\"seq\":0,\"body\":{\"threads\":[{\"name\":\"Thread.Thread(4541273536)\",\"id\":4541273536},{\"nam"+[80]+"\"message\":null,\"success\":1,\"type\":\"response\"}\n") lib/modules/Stdio.pmod/module.pmod:2239: Stdio.FILE(2, "w", 777 /* fd=2 */)->write("\nSending response: \nContent-Length: 228\r\n\r\n{\"request_seq\":16,\"seq\":0,\"b"+[180]+"1,\"type\":\"response\"}\n","Content-Length: 228\r\n\r\n{\"request_seq\":16,\"seq\":0,\"body\":{\"threads\":[{\"n"+[159]+":1,\"type\":\"response\"}") build/darwin-18.7.0-x86_64/master.pike:6576: werror("\nSending response: \n%s\n","Content-Length: 228\r\n\r\n{\"request_seq\":16,\"seq\":0,\"body\":{\"threads\":[{\"name\":\"Thread.Thread(4541273536)\",\"id\":4541273536},{\"na"+[87]+"ge\":null,\"success\":1,\"type\":\"response\"}") lib/modules/Debug.pmod/Debugger.pmod/DebugServer.pike:31: Debug.Debugger.DebugServer()->write_response(Debug.Debugger.DebugAdapterProtocol.ThreadsResponse()) lib/modules/Debug.pmod/Debugger.pmod/DebugServer.pike:145: Debug.Debugger.DebugServer()->handle_threads_request(mapping[3]) lib/modules/Debug.pmod/Debugger.pmod/DebugServer.pike:678: Debug.Debugger.DebugServer()->breakpoint_read(Stdio.File("socket", "127.0.0.1 51892", 777 /* fd=10 */),"Content-Length: 47\r\n\r\n{\"command\":\"threads\",\"type\":\"request\",\"seq\":16}") lib/modules/Stdio.pmod/module.pmod:1165: Stdio.File("socket", "127.0.0.1 51892", 777 /* fd=10 */)->__stdio_read_callback() -:1: Pike.Backend(1)->`()(1000.0) lib/modules/Debug.pmod/Debugger.pmod/DebugServer.pike:540: Debug.Debugger.DebugServer()->__lambda_65700_0_line_540() Abort trap: 6

2 2

Programs with multiple offsets per line
by H William Welliver 12 Feb '21

12 Feb '21

I seem to recall a discussion a year or two ago regarding the fact that a program may have multiple offsets per line (at the time, the breakpoint code I’d written only used the first offset). What scenarios might result in multiple offsets being associated with a single line of code? I’ve tried a few things that would seem obvious (multiple statements, etc), but haven’t been able to detect the situation using the existing code (even when it’s been adjusted to look for it). Bill

1 0

enums
by Lance Dillon 04 Feb '21

04 Feb '21

What would be the best way to represent enums? It has been a while since I wrote pike code, so I'm looking in the docs and can't see anything. Maybe just a bunch of constants? Or maybe a class (or struct)? Thanks

2 2

Recent Pike 8.0 fixes to master and a xenofarm feature request
by H William Welliver 28 Jan '21

28 Jan '21

Hello all- I’ve been working on moving a number of fixes I made during the recent Pike 8.0 release process into master. I’ve tested them all locally, but I’m going to push them so that I can make sure nothing breaks in xenofarm. If you happen to discover breakage before I do, just let me know and I’ll get it squared away. As a side note, I was thinking that it might be nice to display on an individual build result screen (or maybe even on the grid?) the number of failed tests vs tests run. I realize that doesn’t buy you much when you’re just pushing failures from one test to another, but it would make it easy to spot big breakage. This is especially acute on systems that are chronically yellow. Does anyone have any opinions about that? Bill

1 0

Pike 8.0.1116 for Windows (finally!)
by H William Welliver 22 Jan '21

22 Jan '21

Hello all- I’ve spent the past few weeks getting a reasonably full-featured pike build for Windows prepared. I’d be pleased to have folks test this build out and let me know if there are any rough edges. The MSI installer is available at: http://pike.lysator.liu.se/download/pub/pike/all/8.0.1116/ If I don’t hear of any major breakage in the next 3-4 days, I’ll publish an official announcement of the release. Some notes about this build: 1. I’ve used VC2015 on Windows 7, and have not tested the results on WinXP, should anyone still be so fashionably outdated. 2. I did prepare a few newer versions of dependencies (gmp, nettle, pcre, freetype). 3. I managed to find a copy of wix 2.0 on an old VM, and was thus able to build the installer without rewriting a lot of things. It’s pretty much impossible to find a copy of wix2, so fixing the wix input to accommodate the changes for wix3 might be needed in the future. 4. My libs directory for win32 has copies of libraries of UI-ish things, but the build didn’t seem to pick them up automatically, and I don’t really care about them. I don’t know if previous 8.0 releases included support for these things. If anyone does care about GTK or SDL or such, get in touch with me and maybe we can work something out. 5. Most of the problems with the build were related to getting the wix2 tools and various libraries squared away. I did not need to meddle with the source directly, which might be a first. 6. I would be happy to provide all of the details and software required to set up a build environment should anyone else wish to help maintain this functionality. Features: auto_bignum machine_code dynamic_modules threads out-of-band_data COM Crypto.AES.GCM Gettext Gmp Gz HTTPAccept Image.FreeType Image.GIF Image.JPEG Image.PNG Image.TIFF Image.XFace Java Java.NATIVE_METHODS MIME Mysql Nettle PDF.PDFlib Regexp.PCRE Regexp.PCRE.Widestring SQLite Standards.JSON WhiteFish As always, comments and suggestions are welcome! Bill

1 0

QR codes in pike generating a PNG
by Stephen R. van den Berg 18 Jan '21

18 Jan '21

Has anyone implemented a library to generate QR-codes in Pike throwing out a PNG of some kind? -- Stephen.

2 1

8.0 test suite failures: fork() + time on macOS 10.12+
by H William Welliver 05 Jan '21

05 Jan '21

I’m running builds of 8.0 to make sure we don’t have any major test failures, and I’ve run into a few problems so far. I’ll put them in separate emails so they are more manageable. If anyone can offer any assistance, that would be most appreciated. I can supply any info needed, up to getting you a logon to the systems in question. First up, macOS 10.12+ hang on socktest.pike. The 10.11 and earlier do not have this problem, and I haven’t tried running an older binary on a newer OS. The call to gc() in finish() never returns, and according to LLDB: (lldb) thread list Process 4746 stopped * thread #1: tid = 0xa7cc2d, 0x00007fff781f922a libsystem_kernel.dylib`mach_msg_trap + 10, queue = 'com.apple.main-thread', stop reason = signal SIGSTOP (lldb) thread backtrace * thread #1, queue = 'com.apple.main-thread', stop reason = signal SIGSTOP * frame #0: 0x00007fff781f922a libsystem_kernel.dylib`mach_msg_trap + 10 frame #1: 0x00007fff781f976c libsystem_kernel.dylib`mach_msg + 60 frame #2: 0x00007fff781fd05e libsystem_kernel.dylib`clock_get_time + 85 frame #3: 0x0000000105974667 pike`mach_clock_get_time at rusage.c:633:7 frame #4: 0x00000001058e713b pike`do_gc(ignored_UNUSED=<unavailable>, explicit_call=<unavailable>) at gc.c:3507:24 frame #5: 0x00000001059ee723 pike`f_gc(args=<unavailable>) at builtin_functions.c:5126:11 frame #6: 0x00000001063607b7 frame #7: 0x00000001058a365a pike`mega_apply [inlined] eval_instruction(pc=<unavailable>) at interpret.c:1711:5 frame #8: 0x00000001058a3658 pike`mega_apply(type=<unavailable>, args=<unavailable>, arg1=<unavailable>, arg2=<unavailable>) at interpret.c:2695 frame #9: 0x000000010589c72d pike`apply_svalue(s=<unavailable>, args=<unavailable>) at interpret.c:3158:5 frame #10: 0x0000000105a31769 pike`got_fd_event(box=0x0000000105de5008, event=937461904) at file.c:368:5 frame #11: 0x00000001058c901f pike`backend_call_active_callbacks(fd_list=0x00007ffeea373ae8, me_UNUSED=<unavailable>) at backend.cmod:2349:6 frame #12: 0x00000001058c4839 pike`pdb_low_backend_once(pdb=0x00007fef37e086d0, timeout=0x00007ffeea373fa8) at backend.cmod:4137:11 frame #13: 0x00000001058c4aec pike`f_PollDeviceBackend_cq__backtick_28_29(args=1) at backend.cmod:4315:5 frame #14: 0x00000001058a1cbc pike`low_mega_apply(type=APPLY_SVALUE, args=1, arg1=<unavailable>, arg2=<unavailable>) at apply_low.h:221:2 frame #15: 0x00000001058a2753 pike`jump_opcode_F_CALL_FUNCTION_AND_POP at interpret_functions.h:2452:1 frame #16: 0x00000001061d4348 Some other information I discovered looking into this: I tried to add some sleep() in the child process in order to examine the process with dtrace, but the sleep() never returned. Sleep seems to work fine with a pike -e ’sleep(5);’. If I disable the fork(), the test runs successfully. The following test case demonstrates the problem. The sleep() can be exchanged for gc() and it also hangs. int main() { object pid; if (mixed err = catch { pid = fork(); }) { werror("fork() failed\n"); } else if (pid) { int res = pid->wait(); werror("child exited.\n"); return 0; } werror("child\n"); sleep(2); werror("slept\n"); return 0; } bin/pike test2.pike child I don’t quite understand why clock_get_time() would hang like that unless there was some sort of problem with the mach clock service across processes, though it wouldn’t surprise me if that were a problem. What is also interesting is that clock_gettime() is available in 10.12 and newer. According to the manpage, this is POSIX compliant and provides CLOCK_MONOTONIC, which is what is used on some other systems. There is a problem in that _POSIX_MONOTONIC_CLOCK is set to -1, which seems to contradict the man page. Not sure if it makes sense to try that instead, or re-initialize the clock service after the fork? Bill

2 2

WiX tools
by H William Welliver 01 Jan '21

01 Jan '21

I’m working on preparing a windows build, and the version of installer tools available for download appears to be incompatible with the input files for the Pike installer. I suspect that I need WiX 2.0, but it doesn’t seem to be available for download anywhere. Does anyone happen to have a copy of the installer or build environment that could be used to create the installer? The build went surprisingly well (aside from the usual hassle of getting libraries compatible with the version of VS in use). Thanks! Bill

1 0

Pike 8.0.1116 (release)
by H William Welliver 30 Dec '20

30 Dec '20

I’ve cut a new version of Pike 8.0 that includes the missing change log entry as well as a minor fix that hopefully fixes a compile error in the latest version of macOS. The builds all look good so I’m going to declare this an official stable release. Please feel free to upload binaries; I’m going to start working on a set of macOS releases. Unless something goes horribly wrong in the next 24 hours, I’ll announce the new version in a few days. http://pike.lysator.liu.se/download/pub/pike/all/8.0.1116/ Also, what’s the status of Win32 builds? I think my build environment is a bit out of date relative to the last few releases. Bill

3 6

8.0.1116 tarball
by william＠welliver.org 27 Dec '20

27 Dec '20

All- I've discovered that the source tarball contains a handful of extraneous files: a pair of droppings from patch as well as a pair of temporary files my text editor seems to have made (but which I can't find outside the archive). There's nothing of any note in the files, but I don't think they should be allowed to remain. Does anyone have a preference for how I handle this? The "by the book" method would be for me to run make export again, with all that entails, but I could also be persuaded to just remove the files from the archive and re-upload it if that would be less work for others. Also, might it be a good idea to have the export process check for this sort of thing, or maybe operate from a fresh export of the repository? Not sure what the right approach would be... opinions are welcome. Bill

2 1

← Newer
1
...
7
8
9
10
11
12
13
...
305
Older →

Jump to page:

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

Pike-devel