[fpc-pascal] The keyboard unit
Kevin Lyda
kevin at ie.suberic.net
Sat May 11 19:09:04 CEST 2019
And looking a bit further, I see this has been kind of addressed in r39579:
https://svn.freepascal.org/cgi-bin/viewvc.cgi?view=revision&revision=39579
The fix is imperfect - a process could have a ') ' in it, but it would
be unlikely. postfix and postgres appear to put a closing parens in
their argv[0] but always at the end.
So now I'm curious if a) working off fpc trunk is currently an OK idea
and b) if there are some good docs on doing that?
Kevin
On Sat, May 11, 2019 at 9:46 AM Kevin Lyda <kevin at ie.suberic.net> wrote:
>
> Ralf, thanks! That was the idea, a small easy to test program.
>
> [NOTE: this email was a log of me debugging. If you want to skip to
> the issue: it's because the keyboard unit dies when running under tmux
> because /proc/PID/stat is harder to parse than the keyboard unit code
> assumes.]
>
> My setup is similar but slightly different: Ubuntu 18.04 and just
> using a Makefile and the commandline compiler. Am I missing a flag?
> When I run these:
>
> strace -o argh ./test
> strace -o argh ./test-without-keyboard
>
> diff -u argh*
> --- argh 2019-05-11 08:24:54.015836828 +0100
> +++ argh2 2019-05-11 08:25:21.377836828 +0100
> @@ -9,26 +9,6 @@
> ioctl(1, TCGETS, {B38400 opost isig icanon echo ...}) = 0
> ioctl(2, TCGETS, {B38400 opost isig icanon echo ...}) = 0
> readlink("/proc/self/exe", "/home/kevin/src/sandbox/older/fp"..., 255) = 45
> -getpid() = 2158
> -mmap(NULL, 32768, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
> -1, 0) = 0x7bcbfb894000
> -open("/proc/2158/stat", O_RDONLY|O_LARGEFILE) = 3
> -read(3, "2158 (test) R 2156 2156 14867 34"..., 256) = 256
> -close(3) = 0
> -open("/proc/2156/stat", O_RDONLY|O_LARGEFILE) = 3
> -read(3, "2156 (strace) S 14867 2156 14867"..., 256) = 256
> -close(3) = 0
> -mmap(NULL, 32768, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
> -1, 0) = 0x7bcbfb88c000
> -open("/proc/14867/stat", O_RDONLY|O_LARGEFILE) = 3
> -read(3, "14867 (zsh) S 522 14867 14867 34"..., 256) = 256
> -close(3) = 0
> -open("/proc/522/stat", O_RDONLY|O_LARGEFILE) = 3
> -read(3, "522 (tmux: server) S 131 522 522"..., 256) = 256
> -write(2, "Runtime error 106 at $0000000000"..., 39) = 39
> -write(2, " $000000000042968D\n", 20) = 20
> -ioctl(2, TCGETS, {B38400 opost isig icanon echo ...}) = 0
> -write(2, " $000000000040018C\n", 20) = 20
> -write(2, "\n", 1) = 1
> -munmap(0x7bcbfb88c000, 32768) = 0
> -munmap(0x7bcbfb894000, 32768) = 0
> -exit_group(106) = ?
> -+++ exited with 106 +++
> +write(1, "Hello World.\n", 13) = 13
> +exit_group(0) = ?
> ++++ exited with 0 +++
>
> OK, so it seems to read through the ancestors of its process and dies
> when it reads the /proc/PID/stat of the tmux process. So, what if I
> run it without tmux: strace -o argh3 ./test which runs fine and exits
> cleanly. So what's the difference? diff -u argh{,3}
>
> --- argh 2019-05-11 08:24:54.015836828 +0100
> +++ argh3 2019-05-11 08:40:16.456836828 +0100
> @@ -9,26 +9,31 @@
> ioctl(1, TCGETS, {B38400 opost isig icanon echo ...}) = 0
> ioctl(2, TCGETS, {B38400 opost isig icanon echo ...}) = 0
> readlink("/proc/self/exe", "/home/kevin/src/sandbox/older/fp"..., 255) = 45
> -getpid() = 2158
> -mmap(NULL, 32768, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
> -1, 0) = 0x7bcbfb894000
> -open("/proc/2158/stat", O_RDONLY|O_LARGEFILE) = 3
> -read(3, "2158 (test) R 2156 2156 14867 34"..., 256) = 256
> -close(3) = 0
> -open("/proc/2156/stat", O_RDONLY|O_LARGEFILE) = 3
> -read(3, "2156 (strace) S 14867 2156 14867"..., 256) = 256
> -close(3) = 0
> -mmap(NULL, 32768, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
> -1, 0) = 0x7bcbfb88c000
> -open("/proc/14867/stat", O_RDONLY|O_LARGEFILE) = 3
> -read(3, "14867 (zsh) S 522 14867 14867 34"..., 256) = 256
> -close(3) = 0
> -open("/proc/522/stat", O_RDONLY|O_LARGEFILE) = 3
> -read(3, "522 (tmux: server) S 131 522 522"..., 256) = 256
> -write(2, "Runtime error 106 at $0000000000"..., 39) = 39
> -write(2, " $000000000042968D\n", 20) = 20
> -ioctl(2, TCGETS, {B38400 opost isig icanon echo ...}) = 0
> -write(2, " $000000000040018C\n", 20) = 20
> -write(2, "\n", 1) = 1
> -munmap(0x7bcbfb88c000, 32768) = 0
> -munmap(0x7bcbfb894000, 32768) = 0
> -exit_group(106) = ?
> -+++ exited with 106 +++
> +getpid() = 2719
> +mmap(NULL, 32768, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
> -1, 0) = 0x7f8092c23000
> +open("/proc/2719/stat", O_RDONLY|O_LARGEFILE) = 3
> +read(3, "2719 (test) R 2717 2717 2429 348"..., 256) = 256
> +close(3) = 0
> +open("/proc/2717/stat", O_RDONLY|O_LARGEFILE) = 3
> +read(3, "2717 (strace) S 2429 2717 2429 3"..., 256) = 256
> +close(3) = 0
> +open("/proc/2429/stat", O_RDONLY|O_LARGEFILE) = 3
> +read(3, "2429 (zsh) S 2427 2429 2429 3481"..., 256) = 256
> +close(3) = 0
> +open("/proc/2427/stat", O_RDONLY|O_LARGEFILE) = 3
> +read(3, "2427 (ld-linux-x86-64) S 364 242"..., 256) = 256
> +close(3) = 0
> +open("/etc/timezone", O_RDONLY|O_LARGEFILE) = 3
> +read(3, "Etc/UTC\n", 255) = 8
> +close(3) = 0
> +open("/usr/share/zoneinfo/Etc/UTC", O_RDONLY|O_LARGEFILE) = 3
> +read(3, "TZif2\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\1\0\0\0\1\0\0\0\0"...,
> 2048) = 127
> +mmap(NULL, 32768, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS,
> -1, 0) = 0x7f8092c1b000
> +close(3) = 0
> +time([1557560416]) = 1557560416
> +ioctl(1, TCGETS, {B38400 opost isig icanon echo ...}) = 0
> +write(1, "Hello World.\n", 13) = 13
> +munmap(0x7f8092c1b000, 32768) = 0
> +munmap(0x7f8092c23000, 32768) = 0
> +exit_group(0) = ?
> ++++ exited with 0 +++
>
> So the code that does this is in this:
> /usr/share/fpcsrc/3.0.4/rtl/linux/linuxvcs.pp
>
> I instrumented it a bit
> (https://gist.github.com/b110dd08847fbd54e08e8bef803ac5d9) and I see
> the issue:
>
> /proc/5200/stat: 5200 (kbd-test) R 14867 5200 14867 34821
> /proc/14867/stat: 14867 (zsh) S 522 14867 14867 34821
> /proc/522/stat: 522 (tmux: server) Runtime error 106 at $00000000004006D8
> $00000000004006D8
> $0000000000400967
> $000000000040018C
>
> The problem is that /proc/*/stat is kind of hard to parse as this code
> proves - tmux has rewritten its argv[0] to have a space which breaks
> the parser here. And don't think the ")" is a delimiter as argv[0]
> can have a ")" in it. The format seems to imply it needs to be read
> to the end of the line and parsed backwards:
> https://www.redhat.com/archives/axp-list/2001-January/msg00355.html
>
> In addition ttyname(3) is a POSIX function that would avoid this issue.
>
> Kevin
>
> On Sat, May 11, 2019 at 6:08 AM Ralf Quint <freedos.la at gmail.com> wrote:
> >
> > On 5/10/2019 4:42 PM, Kevin Lyda wrote:
> > > If I remove the uses clause it runs fine. Why is the keyboard unit,
> > > which isn't even being used, causing this error?
> > >
> > > Kevin
> >
> > Well, when I got home a short while ago, I tried that small test program
> > on one of my Linux workstations and it runs just fine...
> >
> > FPC 3.04 as installed with Lazarus 2.02 (ok, installed the 3 .deb
> > packages for fpc, fpc-source and lazarus 2.02 from the lazarus
> > SourceForge repo), compiled and running on Linux Mint 19.2/64, tried
> > both Cinnamon and Mate, on two different hosts...
> >
> > Ralf
> >
> >
> > ---
> > This email has been checked for viruses by Avast antivirus software.
> > https://www.avast.com/antivirus
> >
> > _______________________________________________
> > fpc-pascal maillist - fpc-pascal at lists.freepascal.org
> > http://lists.freepascal.org/cgi-bin/mailman/listinfo/fpc-pascal
>
>
>
> --
> Kevin Lyda
> Galway, Ireland
--
Kevin Lyda
Galway, Ireland
More information about the fpc-pascal
mailing list