[BUG] sandbox: './u-boot -l ' fails

Mon Sep 28 15:42:21 CEST 2020

Hi Heinrich,

On Mon, 28 Sep 2020 at 07:30, Heinrich Schuchardt <xypron.glpk at gmx.de> wrote:
>
> On 28.09.20 15:22, Simon Glass wrote:
> > Hi Heinrich,
> >
> > On Mon, 28 Sep 2020 at 05:31, Heinrich Schuchardt <xypron.glpk at gmx.de> wrote:
> >>
> >> On 28.09.20 06:46, Heinrich Schuchardt wrote:
> >>> Am 28. September 2020 06:24:38 MESZ schrieb Simon Glass <sjg at chromium.org>:
> >>>> Hi Heinrich,
> >>>>
> >>>> On Sat, 19 Sep 2020 at 13:48, Heinrich Schuchardt <xypron.glpk at gmx.de>
> >>>> wrote:
> >>>>>
> >>>>> Hello Simon,
> >>>>>
> >>>>> when I try to run ./u-boot -l the sandbox stalls. Shouldn't it run
> >>>> out
> >>>>> of the box?
> >>>>>
> >>>>> $ ./u-boot -l -d arch/sandbox/dts/sandbox.dtb
> >>>>
> >>>> For the record you should be able to use -D to get the same effect as
> >>>> your -d above.
> >>>>
> >>>>>
> >>>>> U-Boot 2020.10-rc4-00018-g21a10244f9-dirty (Sep 19 2020 - 19:55:39
> >>>> +0200)
> >>>>>
> >>>>> Model: sandbox
> >>>>> DRAM:  128 MiB
> >>>>>
> >>>>> Warning: host_lo MAC addresses don't match:
> >>>>> Address in ROM is               26:4b:ca:6c:98:f4
> >>>>> Address in environment is       00:00:11:22:33:44
> >>>>>
> >>>>> Warning: host_virbr0 MAC addresses don't match:
> >>>>> Address in ROM is               ee:3e:c9:ce:1f:9c
> >>>>> Address in environment is       00:00:11:22:33:45
> >>>>>
> >>>>> Warning: host_docker0 MAC addresses don't match:
> >>>>> Address in ROM is               c2:85:07:7b:9a:18
> >>>>> Address in environment is       00:00:11:22:33:46
> >>>>> WDT:   Not found!
> >>>>> MMC:
> >>>>>
> >>>>> No output after this point.
> >>>>>
> >>>>> The problem also exists with U-Boot v2020.07, v2019.10, v2018.11.
> >>>>>
> >>>>> CONFIG_SANDBOX_SDL=y
> >>>>>
> >>>>> SDL_InitSubSystem() never returns. It is looping somewhere in
> >>>> U-Boot's
> >>>>> __serial_getc(). I wonder how it gets there without returning from
> >>>> the
> >>>>> function.
> >>>>>
> >>>>> I compiled SDL2.cpp from
> >>>>> https://gist.github.com/miguelmartin75/6946310#file-sdl2-cpp-L18
> >>>>> with
> >>>>>
> >>>>> g++ SDL2.cpp -D_REENTRANT -I/usr/include/SDL2 -lSDL2 -lGL -o test
> >>>>>
> >>>>> and it runs fine showing an X11 windows with red background.
> >>>>>
> >>>>> So there seems to be no general problem with the SDL2 library.
> >>>>
> >>>> I hit this myself on another computer and it turned out to be that SDL
> >>>> defined getc(), as does U-Boot, and things get confused. At least I
> >>>> think it is getc.
> >>>>
> >>>> I hacked around with changing the name of getc (I think it was getc)
> >>>> in U-Boot and the problem went away.
> >>>
> >>> Should we include a patched SDL2 with U-Boot for static linking?
> >>>
> >>
> >> I cannot find a symbol getc() nor a reference to it in the SDL
> >> repository. getc() is defined in stdio.h.
> >>
> >> Our getc() takes no argument, while the stdio one wants a FILE *.
> >>
> >> But changing the U-Boot definition to "int getc(void *)" does not solve
> >> the issue.
> >
> > I just tried it on the machine where it doesn't work:
> >
> > $ gdb --args /tmp/b/sandbox/u-boot -l -D
> > GNU gdb (Debian 9.2-1) 9.2
> > Copyright (C) 2020 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-linux-gnu".
> > Type "show configuration" for configuration details.
> > For bug reporting instructions, please see:
> > <http://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"...
> > Reading symbols from /tmp/b/sandbox/u-boot...
> > (gdb) br getc
> > Breakpoint 1 at 0x5b6d6: getc. (2 locations)
> > (gdb) r
> > Starting program: /tmp/b/sandbox/u-boot -l -D
> > [Thread debugging using libthread_db enabled]
> > Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1".
> >
> >
> > U-Boot 2020.10-rc5-00196-g0ac83d080a0 (Sep 28 2020 - 07:11:52 -0600)
> >
> > Model: sandbox
> > DRAM:  128 MiB
> >
> > Warning: host_lo MAC addresses don't match:
> > Address in ROM is 1a:34:9b:48:aa:53
> > Address in environment is 00:00:11:22:33:44
> > WDT:   Not found!
> > MMC:
> >
> > Breakpoint 1, getc () at
> > /home/sjg/cosarm/src/third_party/u-boot/files/common/console.c:414
> > 414 if (!gd->have_console)
> > (gdb) up
> > #1  0x00007ffff7914d48 in ?? () from /lib/x86_64-linux-gnu/libX11.so.6
> > (gdb)
> >
> >
> > So it seems that it is libX11 causing the problem.
> >
> > I don't think the best fix is to remove getc() from that library,
> > although I do wonder if it is a bug. Renaming the symbol in U-Boot
> > with #define (only on sandbox) might be one option.
> >
> > Regards,
> > Simon
> >
>
> #define getc _uboot_getc
>
> is what I tried but it runs you into the trouble that many other symbols
> contain the getc substring.
>
> As we always try to use the same definition in U-Boot as in glibc we
> should really replace getc() by another symbol, e.g. chget() so that we
> avoid confusion.

Well the first question is whether this is a bug with x11.

I'm not sure that renaming is a good idea. getc() is a pretty standard name.

Try

#define getc(x) _sandbox_get(x)

which might get you closer.

Regards,
Simon