[Qemu-devel] Fastest TCP/IP Stack
I'm running a terminal server under qemu with kqemu compiled into my kernel under the -kernel-kqemu for fastest performance. What is the most efficient method of -net ? I was using -net user with OpenVPN to connect to my internal LAN, but I have switched to -net tap to see if that is faster. Does anyone have any knowledge of which would take the least overhead? I noticed on my top stats that my % sys was particularly high when using networking under -net user. Thanks -Joseph ___ Qemu-devel mailing list Qemu-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/qemu-devel
Re: [Qemu-devel] Fastest TCP/IP Stack
I've not done any benchmarks, but I do pay attention to resource usage. (Host OS linux) -net user has pretty good thoughput, but does burn some cpu. I think in that regard, -net tap has less overhead. During my use of qemu, I've noticed that the quality of the nic driver and the nic emulation itself comes into play. I would try the differnet models and see. -net nic,model=pcnet always seemed like it performed well. YOMV. WD On 9/14/06, Joseph Miller [EMAIL PROTECTED] wrote: I'm running a terminal server under qemu with kqemu compiled into my kernel under the -kernel-kqemu for fastest performance. What is the most efficient method of -net ? I was using -net user with OpenVPN to connect to my internal LAN, but I have switched to -net tap to see if that is faster. Does anyone have any knowledge of which would take the least overhead? I noticed on my top stats that my % sys was particularly high when using networking under -net user. Thanks -Joseph ___ Qemu-devel mailing list Qemu-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/qemu-devel -- 22:38 @WaxDragon false ^ true 22:39 false :( 22:39 false dont you think you can XOR me and get away with it! I always return! ___ Qemu-devel mailing list Qemu-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/qemu-devel
Re: [Qemu-devel] Fastest TCP/IP Stack
On 9/14/06, Joseph Miller [EMAIL PROTECTED] wrote: I'm running a terminal server under qemu with kqemu compiled into my kernel under the -kernel-kqemu for fastest performance. What is the most efficient method of -net ? I was using -net user with OpenVPN to connect to my internal LAN, but I have switched to -net tap to see if that is faster. Does anyone have any knowledge of which would take the least overhead? I noticed on my top stats that my % sys was particularly high when using networking under -net user. Thanks -net tap is considerably more efficient than -net user in most cases. -net user is a TCP/UDP proxy, so (a) your TCP connections are no longer end-to-end, and (b) performance is at the mercy of qemu's internal TCP/IP stack. With -net tap and a bridge, IP packets are passed along unmolested, and all the magic occurs within the host kernel. (Of course where performance is not a pressing issue, -net user is still awfully convenient.) --Ed ___ Qemu-devel mailing list Qemu-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/qemu-devel
[Qemu-devel] cpustate patch
While hunting a tls related bug in ARM emulation, I found that that qemu distributed with scratchbox includes the attached patch. Without the patch python's configure halts when checking for the -pthread option. The patch inverts the order of the list and makes the code simpler. I think that there is a remaining problem, if first_cpu == NULL then env-cpu_index is not initialized. Should an else be added? Any thoughts on the patch? Thanks, Rafael cpustate.patch Description: Binary data ___ Qemu-devel mailing list Qemu-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/qemu-devel
Re: [Qemu-devel] cpustate patch
Rafael Espíndola wrote: While hunting a tls related bug in ARM emulation, I found that that qemu distributed with scratchbox includes the attached patch. Without the patch python's configure halts when checking for the -pthread option. The patch inverts the order of the list and makes the code simpler. I think that there is a remaining problem, if first_cpu == NULL then env-cpu_index is not initialized. Should an else be added? Any thoughts on the patch? For consistency I prefer that the first cpu of the list is number zero, so I prefer to leave the code as it is now. Regards, Fabrice. ___ Qemu-devel mailing list Qemu-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/qemu-devel
Re: [Qemu-devel] PCI IRQ routing problems
Ed Swierk wrote: Linux 2.6.17 running on the latest qemu snapshot is unable to route IRQs to more than 4 network interfaces when running without ACPI, and is limited to 2 network interfaces with ACPI enabled. [...] I suspect the problem in the non-ACPI case is caused by a limitation in the PCI IRQ routing table in the Bochs BIOS, but I haven't a clue how to fix it. Any ideas would be appreciated. I did not study the issue yet, but I suspect it is a Bochs BIOS issue (I remember it declares a limited number of PCI slots for example). IMHO, all the PCI, MP table and ACPI table init should be moved to the Bochs BIOS. I did not do it because it was not strictly necessary to have something that works. A simple way to do that is to copy the relevant QEMU code in a separate section of the Bochs BIOS and to execute it by switching temporarily to 32 bit protected mode in the BIOS init code. Fabrice. ___ Qemu-devel mailing list Qemu-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/qemu-devel
[Qemu-devel] OSX 10.4.7 build
I am trying to build qemu-cvs on OSX 10.4.7, and I am getting the following error: gcc -DQEMU_TOOL -Wall -O2 -g -fno-strict-aliasing -I. -mdynamic-no-pic -g -D_GNU_SOURCE -D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -o qemu-img qemu-img.c block.c block-raw.c block-cow.c block-qcow.c aes.c block-vmdk.c block-cloop.c block-dmg.c block-bochs.c block-vpc.c block-vvfat.c block-qcow2.c -lz block.c: In function 'bdrv_commit': block.c:448: error: 'ENOMEDIUM' undeclared (first use in this function) -- -- Troy Benjegerdes'da hozer'[EMAIL PROTECTED] Somone asked me why I work on this free (http://www.fsf.org/philosophy/) software stuff and not get a real job. Charles Shultz had the best answer: Why do musicians compose symphonies and poets write poems? They do it because life wouldn't have any meaning for them if they didn't. That's why I draw cartoons. It's my life. -- Charles Shultz ___ Qemu-devel mailing list Qemu-devel@nongnu.org http://lists.nongnu.org/mailman/listinfo/qemu-devel
