[Qemu-devel] Bugs

[malc]

Following is the list of some easy to reproduce bugs in QEMU,
perhaps someone will have motivation to fix them:

Preliminary:

cvs -d :ext:address@hidden:/cvsroot/qemu login
cvs -d :ext:address@hidden:/cvsroot/qemu co qemu
cd qemu
./configure --target-list=i386-softmmu
make

======================================================================
DR-DOS EMS

Steps:
$ wget http://www.boblycat.org/~malc/drdos.fda.bz2
$ bunzip2 drdos.fda.bz2
$ i386-softmmu/qemu -L pc-bios -fda drdos.fda

[enter]
[enter]
C:\>wrms

Under other circumstances (without EMS driver, or on MS-DOS 6.2 with
or without EMS driver) this thing should just print an error message
and exit. With DR-DOSes EMS driver it enters endless exception loop.

This issue probably will arise for many non trivial DOS4G/W programs.

======================================================================
Hard Rox demo

Steps:
$ wget http://www.boblycat.org/~malc/rox.hda.bz2
$ bunzip2 rox.hda.bz2
$ i386-softmmu/qemu -L pc-bios -hda rox.hda

[enter]
[enter]
C:\>cd rox
C:\ROX>rox

Again an endless do_interrupt loop but this time it's more severe
since QEMU never gets around of checking SDL event queue and GUI
window is unkillable (by normal means). I would guess it's some
sort of MMU emulation fault.

This issue probably will arise for many non trivial go32 programs.

======================================================================
ReactOS floppy

Steps:
Obtain live CD from:
http://prdownloads.sourceforge.net/reactos/reactos0.2.6-REL-live.zip?download

$ unzip reactos0.2.6-REL-live.zip
$ ./qemu-img create test.fda 1440K
$ gdb --args i386-softmmu/qemu -L pc-bios -cdrom roslive.iso -fda test.fda

Navigate to Start -> Run. In the dialog type: cmd

[Make sure to ungrab the mouse]

C:\reactos>a:

0x00000000 in ?? ()
(gdb) bt
#0  0x00000000 in ?? ()
#1  0x0804b727 in qemu_run_timers (ptimer_head=0x8106440, current_time=The 
value of variable 'current_time' is distributed across several
locations, and GDB cannot access its value.

) at /mnt/big/npf/cvs/qemu/qemu-vanilla/vl.c:760
#2  0x0804dbd0 in main_loop_wait (timeout=0) at 
/mnt/big/npf/cvs/qemu/qemu-vanilla/vl.c:2742
#3  0x0804dcd5 in main_loop () at /mnt/big/npf/cvs/qemu/qemu-vanilla/vl.c:2787
#4  0x0804f33a in main (argc=9, argv=0xbfffefb4) at 
/mnt/big/npf/cvs/qemu/qemu-vanilla/vl.c:3820

With some other QEMU builds (different build options and so on) i get
different back-traces. Something is really fishy here, as if QEMU
manages to really screw up memory in host space. I was getting
segfaults in code_mem_writel and free.

--
mailto:address@hidden