diff options
| author | Rodrigo Arias Mallo <rodarima@gmail.com> | 2025-09-28 20:26:15 +0200 |
|---|---|---|
| committer | Rodrigo <rodarima@gmail.com> | 2025-09-28 23:10:55 +0200 |
| commit | fb510ea86be5ceb9e91573890242581fdbd77ad8 (patch) | |
| tree | d819fe40683592008d136727f5a0b03e48dc1164 /429 | |
Diffstat (limited to '429')
| -rw-r--r-- | 429/index.md | 253 | ||||
| -rw-r--r-- | 429/meta | 4 |
2 files changed, 257 insertions, 0 deletions
diff --git a/429/index.md b/429/index.md new file mode 100644 index 0000000..2efb7e9 --- /dev/null +++ b/429/index.md @@ -0,0 +1,253 @@ +Title: Dillo 3.2.0: view source consistently freezes +Author: mbuechse +Created: Mon, 11 Aug 2025 10:29:45 +0000 +State: closed + +On Alpine: + +```shell +$ dillo file:/usr/share/doc/dillo/user_help.html +paths: Cannot open file '/home/mbue/.dillo/keysrc': No such file or directory +paths: Using /etc/dillo/keysrc +paths: Cannot open file '/home/mbue/.dillo/domainrc': No such file or directory +paths: Using /etc/dillo/domainrc +Domain: Default accept. +dillo_dns_init: Here we go! (threaded) +TLS library: OpenSSL 3.5.1 1 Jul 2025 +Disabling cookies. +paths: Cannot open file '/home/mbue/.dillo/hsts_preload': No such file or directory +paths: Using /etc/dillo/hsts_preload +Nav_open_url: new url='file:/usr/share/doc/dillo/user_help.html' +Nav_open_url: new url='dpi:/vsource/:file:/usr/share/doc/dillo/user_help.html' +^C + +$ ps -elf | grep vsource + 8695 mbue 0:06 /usr/lib/dillo/dpi/vsource/vsource.filter.dpi + 8727 mbue 0:00 grep vsource + +$ kill 8695 +``` + +The filter runs at a high CPU usage, causing my laptop's fan to spin up, but it doesn't produce anything. + +--%-- +From: rodarima +Date: Mon, 11 Aug 2025 11:34:02 +0000 + +Thanks for the report, but I cannot reproduce it here. Can you record a trace of the `vsource.filter.dpi` program with strace during the issue? Something like: + +``` +$ strace -o strace.log -s 500 -p $(pidof vsource.filter.dpi) +(wait a few seconds, then Ctrl+C to stop) +``` + +Then attach the strace.log file here. + +--%-- +From: mbuechse +Date: Mon, 11 Aug 2025 12:16:26 +0000 + +Thanks for the precise instructions! I'm afraid the file came out empty. +I could only start strace after I had opened the source view, because otherwise there was no process to attach to. +And then, apparently, nothing happened. +I will give it more time (longer than a few seconds). + +--%-- +From: mbuechse +Date: Mon, 11 Aug 2025 12:19:40 +0000 + +A quick check with `gdb` + +```shell +# gdb -p $(pidof vsource.filter.dpi) +GNU gdb (GDB) 15.2 +Copyright (C) 2024 Free Software Foundation, Inc. +License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> +This is free software: you are free to change and redistribute it. +There is NO WARRANTY, to the extent permitted by law. +Type "show copying" and "show warranty" for details. +This GDB was configured as "x86_64-alpine-linux-musl". +Type "show configuration" for configuration details. +For bug reporting instructions, please see: +<https://www.gnu.org/software/gdb/bugs/>. +Find the GDB manual and other documentation resources online at: + <http://www.gnu.org/software/gdb/documentation/>. + +For help, type "help". +Type "apropos word" to search for commands related to "word". +Attaching to process 18067 +Reading symbols from /usr/lib/dillo/dpi/vsource/vsource.filter.dpi... +(No debugging symbols found in /usr/lib/dillo/dpi/vsource/vsource.filter.dpi) +Reading symbols from /lib/ld-musl-x86_64.so.1... +Reading symbols from /usr/lib/debug//lib/ld-musl-x86_64.so.1.debug... +printf_core (f=f@entry=0x7ffd23d137e0, + fmt=fmt@entry=0x55ad401ec073 "\n<html><head>\n<title>Source for %s</title>\n<style type=\"text/css\">\n body {\n white-space: pre-wrap;\n font-family: monospace;\n margin: 0;\n width: 100%;\n }\n table { border:0 }\n td.num {\n "..., ap=ap@entry=0x7ffd23d13648, + nl_arg=nl_arg@entry=0x7ffd23d136e0, nl_type=nl_type@entry=0x7ffd23d13660) + at src/stdio/vfprintf.c:457 + +warning: 457 src/stdio/vfprintf.c: No such file or directory +``` + +--%-- +From: rodarima +Date: Mon, 11 Aug 2025 12:20:58 +0000 + +Thanks!, can you share the backtrace with `bt`? + +--%-- +From: mbuechse +Date: Mon, 11 Aug 2025 12:26:29 +0000 + +Sure. It doesn't tell much, I'm afraid. + +```text +(gdb) bt +#0 printf_core (f=f@entry=0x7ffd23d137e0, + fmt=fmt@entry=0x55ad401ec073 "\n<html><head>\n<title>Source for %s</title>\n<style type=\"text/css\">\n body {\n white-space: pre-wrap;\n font-family: monospace;\n margin: 0;\n width: 100%;\n }\n table { border:0 }\n td.num {\n "..., ap=ap@entry=0x7ffd23d13648, + nl_arg=nl_arg@entry=0x7ffd23d136e0, nl_type=nl_type@entry=0x7ffd23d13660) + at src/stdio/vfprintf.c:457 +#1 0x00007f5de41c9100 in vfprintf (f=f@entry=0x7ffd23d137e0, + fmt=0x55ad401ec073 "\n<html><head>\n<title>Source for %s</title>\n<style type=\"text/css\">\n body {\n white-space: pre-wrap;\n font-family: monospace;\n margin: 0;\n width: 100%;\n }\n table { border:0 }\n td.num {\n "..., ap=<optimized out>) at src/stdio/vfprintf.c:690 +#2 0x00007f5de41c95af in vsnprintf (s=<optimized out>, n=<optimized out>, + fmt=<optimized out>, ap=<optimized out>) at src/stdio/vsnprintf.c:49 +#3 0x000055ad401eae98 in ?? () +#4 0x000055ad401eafb5 in ?? () +#5 0x000055ad401e9673 in ?? () +#6 0x000055ad401e91ac in ?? () +#7 0x00007f5de41a7496 in libc_start_main_stage2 (main=0x55ad401e9030, + argc=1, argv=0x7ffd23d13b78) at src/env/__libc_start_main.c:95 +#8 0x000055ad401e925d in ?? () +#9 0x0000000000000001 in ?? () +#10 0x00007ffd23d14beb in ?? () +#11 0x0000000000000000 in ?? () +``` + +--%-- +From: rodarima +Date: Mon, 11 Aug 2025 12:31:32 +0000 + +Based on the fact that you are not walking through any syscall, it looks that you may be looping here: + +https://github.com/dillo-browser/dillo/blob/v3.2.0/dlib/dlib.c#L407-L432 + +There may be a corner case for that particular size of the buffer that is causing it to never be able to find a suitable size for that string. + +Are you able to view the source code of other pages which have a different title? For example `about:splash`? + +--%-- +From: mbuechse +Date: Mon, 11 Aug 2025 12:40:54 +0000 + +No, other pages don't work either. Maybe it's connected to musl? +https://git.musl-libc.org/cgit/musl/tree/src/stdio/vsnprintf.c?h=v1.2.5 + +--%-- +From: rodarima +Date: Mon, 11 Aug 2025 13:16:25 +0000 + +Maybe. We can confirm that it is the case by adding these debug statements: + +```diff +diff --git a/dlib/dlib.c b/dlib/dlib.c +index 2cbd083e..801d8abc 100644 +--- a/dlib/dlib.c ++++ b/dlib/dlib.c +@@ -402,6 +402,8 @@ void dStr_vsprintfa (Dstr *ds, const char *format, va_list argp) + { + int n, n_sz; + ++ fprintf(stderr, "dStr_vprintfa: enter ds->len=%d, ds->sz=%d\n", ds->len, ds->sz); ++ + if (ds && format) { + va_list argp2; /* Needed in case of looping on non-32bit arch */ + while (1) { +@@ -428,9 +430,13 @@ void dStr_vsprintfa (Dstr *ds, const char *format, va_list argp) + n_sz = ds->sz * 2; + } + #endif ++ fprintf(stderr, "dStr_vprintfa: resizing, n=%d, n_sz=%d, keep=%d\n", ++ n, n_sz, (ds->len > 0) ? 1 : 0); + dStr_resize(ds, n_sz, (ds->len > 0) ? 1 : 0); + } + } ++ ++ fprintf(stderr, "dStr_vprintfa: exit ds->len=%d, ds->sz=%d\n", ds->len, ds->sz); + } + + /** +``` + +[debug-dlib.patch.txt](https://github.com/user-attachments/files/21715372/debug-dlib.patch.txt) + +You will need *first remove the current dillo 3.2.0* and then build it from source. I suggest trying with the tip of master, as it should be reproducible there: + + +``` +git clone https://github.com/dillo-browser/dillo.git +cd dillo +git apply < debug-dlib.patch.txt +./autogen.sh +mkdir build +cd build +../configure CFLAGS='-Og -g' CXXFLAGS='-Og -g' +make +doas make install # must be installed! +dpidc stop +``` + +Then try again, if it fails, it will start looping on the "dStr_vprintfa: resizing" line. We will see also which size parameter is causing it to loop. + +--%-- +From: rodarima +Date: Mon, 11 Aug 2025 13:31:49 +0000 + +Oh, I see what may be going on: + +https://github.com/dillo-browser/dillo/blob/b88506f619950640a7ea0dc3a1f615dc71068674/dpi/vsource.c#L127 + +That should be `%%` as otherwise is a printf format. It must be confusing musl. + +--%-- +From: mbuechse +Date: Mon, 11 Aug 2025 13:33:56 +0000 + +> That should be %% as otherwise is a printf format + +I won't be able to build and test until the evening (in 5 hrs maybe). Then I could test this hypothesis as well, while I'm at it. + +--%-- +From: rodarima +Date: Mon, 11 Aug 2025 13:38:57 +0000 + +Thanks, I tested this on my old RPI 2 with Alpine and it seems to be the case: + +``` +berry:~$ cat a.c +#include <stdio.h> + +int main() +{ + char buf[16]; + const char *fmt = "oops%;"; + int n = snprintf(buf, 16, fmt, "trash"); + printf("n=%d\n", n); + + return 0; +} +berry:~$ gcc a.c -o a +berry:~$ ldd ./a + /lib/ld-musl-armhf.so.1 (0x76f72000) + libc.musl-armv7.so.1 => /lib/ld-musl-armhf.so.1 (0x76f72000) +berry:~$ ./a +n=-1 +``` + +It always fails with -1, so it assumes that is the old glibc behavior so it keeps expanding the buffer thinking that the problem is that it is not big enough. + +I'll open a PR shortly. + +--%-- +From: mbuechse +Date: Mon, 11 Aug 2025 18:50:10 +0000 + +That was really quick. Thank you very much!
\ No newline at end of file diff --git a/429/meta b/429/meta new file mode 100644 index 0000000..1b9c8bd --- /dev/null +++ b/429/meta @@ -0,0 +1,4 @@ +title="Dillo 3.2.0: view source consistently freezes" +state=Closed +created="Aug 11, 2025, 12:29 PM GMT+2" +author=mbuechse |