hang when probing vbe modes

[stsp]

I tried
sudo hwinfo --vbe
and it hanged on probing video modes.

(gdb) bt
#0 0x00007fbecd31292a in tsc () at include/x86emu_int.h:121
#1 x86emu_run (emu=0x5803519f, flags=25) at decode.c:252
#2 0x000000000207c8d0 in ?? ()
#3 0x00007fbecd9141b1 in vm_run (emu=0x20cc0f0, emu@entry=0x2212a80,

t=0x2212a80, t@entry=0x7ffc46dd1948) at mdt.c:327

#4 0x00007fbecd914948 in list_modes (vm=0x110, vm@entry=0x20cc0f0,

vbe=vbe@entry=0x207c8d0) at mdt.c:557

#5 0x00007fbecd915a85 in get_vbe_info (hd_data=0x2079010, vbe=0x207c8d0)

at mdt.c:181

#6 0x00007fbecd9462db in hd_scan_bios (hd_data=hd_data@entry=0x2079010)

at bios.c:451

#7 0x00007fbecd926bc0 in hd_scan_no_hal (hd_data=0x2079010) at hd.c:2027
#8 hd_scan (hd_data=hd_data@entry=0x2079010) at hd.c:1913
#9 0x00007fbecd9277c8 in hd_list (hd_data=0x2079010, item=item@entry=hw_vbe,

rescan=rescan@entry=1, hd_old=hd_old@entry=0x0) at hd.c:3235

#10 0x0000000000404387 in do_hw (hd_data=hd_data@entry=0x2079010,

f=f@entry=0x0, hw_item=50) at hwinfo.c:548

#11 0x00000000004026f6 in main (argc=2, argv=) at hwinfo.c:365

I've googled out the similar problem here:
http://bugs.rosalinux.ru/show_bug.cgi?id=7299

Unfortunately, while trying to debug it, it have suddenly disappeared.
I guess perhaps something is not properly initialized?

[stsp]

After reboot, the problem reliably re-appears.
So what it does is some I/O in PCI config space
in a loop.
Maybe the mode listing should simply be disabled?
Not the most important info IMHO.

[stsp]

No, not PCI conf.
It talks to video card regs at 0xe000 IO space.

[stsp]

It turns out the problem happens only if booted
with KMS driver. With nomodeset=1 the problem
does not happen. This may be a vbios quirk, but
in general using VBE functions with kms is unsafe.
I suspect the mode listing should be disabled.
Hope at least 4f00 func is safe as it doesn't need
to talk to card perhaps.

[wfeldt]

Please test the latest version.