| |
| |
| |
|
Page: 1 2 3
Comments:
<0> can i run a 32bit os inside qemu-system-x86_64
<1> yep
<0> vista doesnt like it too much
<0> BSOD on me
<2> a known problem. someone was working to have it running on the qemu-devel list.
<2> involved some changes to the acpi code ?
<0> wouldnt be surprised
<0> dignome, which 32bit os do you know that work under qemu-system-x86_64
<0> do i have access to the sse2 and sse3 extensions under qemu?
<2> many. linux, windows (obviously not vista), bsd, solaris, etc...
<0> QNX doesnt like it
<0> only works with qemu, not with -system-x86_64
<2> if you do qemu-system-x86_64 -no-kqemu ... does it work?
<2> kqemu makes ***umptions as to how guest os should run so not many will run with it.
<0> yes
<0> wats with that?
<2> do you have a Intel VT or AMD-V capable processor? if so could try KVM.
<3> ths: if you find some time, could you please also have a look to the patch to fix the issue withe GT64120
<3> .. the GT64120 and REDBOOT
<2> cruiseoveride: no idea what kqemu does internally. only Fabrice knows all the semantics.
<0> <2> i have common Athlon X2
<0> does that have VT?
<4> hm... quick question.. i've started sshd and i know the ip of the virtual box but i cannot connect,... can someone hint me were to look to set up this.. iptables is not supported in the box
<5> aurel32: I first wanted to fix 64bit malta, so I can test easier.
<2> cruiseoveride: grep svm /proc/cpuinfo
<6> .oO( wasn't that "crash override"? )
<4> the virtualbox cannot be pinged..
<3> ths: ok I see
<2> azi`: the way slirp works you have to use -redir to get in from outside the guest
<0> <2> i got vme, no svm
<2> k
<4> dignome: slrip? what's that?
<4> slirp*
<2> azi`: that is what qemu uses for networking by default -> http://qemu.org/qemu-doc.html#SEC23
<4> thanks
<0> does anyone know the link of all working OS under qemu?
<2> cruiseoveride: yes. it's filled with spam though - still want to see it?
<0> wats with this guy Fabrice , does he think hes gonna make money off kqemu ?
<0> yah
<2> cruiseoveride: http://www.claunia.com/qemu
<0> wats with the spam
<0> who the hell spams a support list
<0> doesnt anyone moderate this page?
<0> seems some dude has vista beta2 working from cvs,
<0> i got the vista release
<0> i guess it isnt working anymore
<3> ths: you're latest changes to the malta emulation causes a segfault here
<2> cruiseoveride: http://lists.gnu.org/archive/html/qemu-devel/2006-11/msg00255.html - i guess this is that thread mentioned
<0> that dude needs to get the release and play with it
<0> sp1 is also in rc stage for Vista, so its changing fast
<0> how does bosch match up with qemu in terms of speed of running a guest os
<2> cruiseoveride: bochs and speed don't usually go in the same sentence.
<0> im getting tired of this
<2> cruiseoveride: it is more accurate in its emulation - http://bochs.sourceforge.net/
<0> the only reason i have spent about 30hrs worth of playing with qemu is because my vmmon.ko fails to compile, i think with the time spent messing around i could have rewritten the entire vmmon.ko by now
<0> <2> do u know if sse2 and see3 are available under qemu?
<2> it emulates them, yes.
<5> aurel32: Even after a full rebuild?
<3> I am rebuilding now
<3> ths: also I am not sure that the prom_set change is right
<3> the kernel expect to see a 32-bit address even if 64-bit if I am right
<3> well at least it's what I recall from reading the kernel code
<5> It segfaulted in that place for 64bit.
<3> ah ok then
<3> also why to change unsigned to signed
<3> ?
<3> address are unsigned if you want to add them that can makes a difference
<3> adresses
<3> ths: same problem after a full rebuild
<3> ths: I am trying to debug that
<5> aurel32: In order to get the sign extensions without myriads of casts.
<5> On MIPS you can claim addresses are signed.
<0> who know the sign extension instruction on the hc12?
<0> common its easy
<5> cruiseoveride: The CPU manual.
<0> ill give u guys clues
<3> ths: but that causes a problem on 64-bit hosts
<3> ths: switching back to uint32_t fixes the problem
<0> It something that is better when you get it for free, it is also something, that if you want it for free, you also don't control how much you get
<0> <5> indulge me, take a wild guess
<5> aurel32: Ah, missed that.
<3> ths: but maybe my code is also wrong
<0> ah u guys are boring, the sign extension instruction on the hc12 is ***
<3> at some point I mix host addresses and guest addresses
<3> and adding (int32_t)0x80002000 or (uint32_t)0x80002000 make a huge difference
<3> that does not break the r4k code though
<3> I mean mips_r4k
<7> ping ths
<7> hi people
<1> hi there
<5> Usurp: Hello.
<5> aurel32: I changed efectively nothing for the mips_r4k.
<7> qemu complains about not finding qemu.h, any chnages about it ?
<7> qemu/exec.c, sorry
<5> Hm, it doesn't for me.
<3> INITRD_LOAD_ADDR is ok as int32_t
<7> make[1]: Entering directory `/home/syl/qemu/i386-user'
<2> Usurp: which targets are selected to build?
<7> gcc34 -Wall -O2 -g -fno-strict-aliasing -I. -I.. -I/home/syl/qemu/target-i386 -I/home/syl/qemu -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -I/home/syl/qemu/fpu -DHAS_AUDIO -I/home/syl/qemu/slirp -c -o exec.o /home/syl/qemu/exec.c
<7> /home/syl/qemu/exec.c:38:18: qemu.h: No such file or directory
<7> i386-softmmu & i386-user
<3> ENVP_ADDR and VIRT_TO_PHYS_ADDEND seems to cause problems
<5> aurel32: AFAIR I changed INITRD_LOAD_ADDRESS to get a 64bit address print out.
<3> ENVP_ADDR should actually always be 32-bit
<3> even on 64-bit target
<3> look at arch/mips/mips-boards/generic/cmdline.c
<3> * YAMON (32-bit PROM) p*** arguments and environment as 32-bit pointer.
<3> * This macro take care of sign extension.
<5> aurel32: I like to see the sign extension in a 64bit qemu.
<2> Usurp: there was a checkin for darwin qemu user targets - maybe that broke it...
<2> Usurp: i get the same build failure when using --target-list=i386-user
<5> Usurp: Try configuring for i386-linux-user
<2> oh, yeah - forgot ;p
<7> ths - distclean + build complete with only i386-linux-user as target
<5> aurel32: I think INITRD_LOAD_ADDR/ENVP_ADDR need a cast to int32_t before getting widened.
<3> ths: well the problem here is that those #define are only used on the host
<5> Usurp: Well, $foo-linux-user instead of $foo-user.
<3> well for ENVP_ADDR, int32_t is ok
<3> but it should not be int64_t on 64-bit target
<3> int32_t in all cases
<7> ths: this a workaround only, right ?
<7> is*
<5> Usurp: The name of the target changed because there are darwin targets as well now.
<7> ok, which one is correct to emulate an i386 windows-like OS ? (running ReactOS mainly)
<5> Usurp: None of the -user ones. You have to use full system emulation.
<7> then the i386-softmmu is the right one ?
<5> x86-64 may work as well.
<8> it s possible to add support wireless card in qemu boot windows xp ?
<9> sekoica: not sure i follow what you mean
<9> do you want to emulate a wireless card?
<8> i run linux and windows in qemu
<8> yes
<8> for Netstumbler
<7> ths: atm ReactOS is built (here) for an i386 with athlon-xp
<3> ths: AFAIK, kernels, initrd and environment variables are always in kseg0
<3> ths: so I don't know why we need 64-bit variables here
<3> ths: the only point that need some attention is loading the program counter
<3> ths: and AFAIR, on the Malta platform, 64-bit kernels are actually embedded in a 32-bit ELF file
<5> aurel32: Kernel could also be in XKPHYS (they are for e.g. Octane).
<5> The Malta doesn't know about ELF at all, it loads srec files.
<5> I want to keep some consistency: 64bit address space -> 64bit wide addresses.
<5> That said, I still fail to see what goes wrong for you. I have not 64bit host at hand to test.
<3> basically things like that: phys_ram_base + ENVP_ADDR + VIRT_TO_PHYS_ADDEND
<7> thanks, night people
<3> ENVP_ADDR is a negative number so it is actually substracted from phys_ram_base
<5> I see.
<3> VIRT_TO_PHYS_ADDEND is not a problem because it is always 64-bit
<3> it could cause a problem on 128-bit hosts though :)
<3> ths: do you want an access on a 64-bit machine to make some tests ?
<5> aurel32: I figure using 32bit constants (except for the addend) will fix this.
Return to
#qemu
or
Go to some related
logs:
print screen problem xgl
postqueue delete
ubuntu GDM Xserver not found
arptype 0 not supported by libpcap
don't have the right GStreamer
scipy mat class doc
postfix + deffered queue restore
#math
#web
kill bulio
|
|