similar to: PATCH: pretty print for xenperf

(CCed to xen-devel for completeness. ;) On Wed, 2006-08-16 at 17:24 +0200, Tristan Gingold wrote: > I am porting xen-ppc''s xencomm to xen/ia64. > Currently on xen/ia64 copy_from/to_guest uses guest virtual address. This > works well as long as the virtual addresses are in the TLB. When not in TLB > (or vTLB) the hypercall can''t success without domain help. The

Hi, taking into account Hollis comments I now submit this patch. IA64 specific stuff will be posted to xen-ia64-unstable after merge. Tristan. _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com http://lists.xensource.com/xen-devel

Hi, xencomm is the ppc infrastructure to do hypercalls with guest physical addresses instead of virtual address. Because ia64 should also use physicall address, I think it''s better to re-use the ppc code and to put into common code. I''d propose to submit this patch is every part is OK (ie no NAK). This patch creates include/xen/xencomm_access.h which is to be included by

I think this is the last of the header fixes I''ve run across. Though it''s sometimes difficult to tell, I believe Xen/ia64 has asm/mm.h, flushtlb.h, page.h, and shadow.h. Please apply. Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com> -- Hollis Blanchard IBM Linux Technology Center

This patch pretty prints the hypercall section for $xenperf -f Each hypercall count is tagged by its name. Reference: http://lists.xensource.com/archives/html/xen-ia64-devel/2006-08/msg00261.html Signed-off-by Ken Hironaka <kenny@logos.ic.i.u-tokyo.ac.jp> _______________________________________________ Xen-devel mailing list Xen-devel@lists.xensource.com

Hi, on IA64 flushing the whole TLB is very expensive: this is a cpu tlb flush and clearing 16MB of memory (virtual tlb). However, flushing an address range is rather cheap. Flushing an address range on every processors is also cheap (no IPI). Unfortunatly Xen common code flushes the whole TLB after unmapping grant reference. Currently, this is not done on IA64 because domain_dirty_cpumask

Hi Ewan, I''m almost ready to integrate some PPC-specific stuff into xend, and I was wondering if you had a plan for how that should work. First example: the device tree data structure we talked about a few weeks ago. We will need to pass the config data to PPC code, probably in XendDomainInfo.initDomain(), and then pass the resulting data structure into libxc''s xc_linux_load()

# HG changeset patch # User Hollis Blanchard <hollisb@us.ibm.com> # Date 1158780052 18000 # Node ID f7d90f962967a5a94fce0c04f8fcac449f36344f # Parent 041be3f6b38e05f904d240630c18cadb1259317b [XEND] Remove hard tabs. Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com> diff -r 041be3f6b38e -r f7d90f962967 tools/python/xen/xend/XendDomainInfo.py ---

This patch changes the dom0_ops structures as discussed in the thread "32/64-bit hypercall interface". Keir, I added a struct inside XENIF_PTR() to catch direct users in the general code; it was quite useful to have the compiler identify those spots. I have compiled x86_32 and run it with xm-test[1] under qemu. There are 63 passed tests, so that''s good. I still need to

Hi Ewan/Alistair, I have a patch that looks like this: diff -r a39ad4c78850 tools/libxc/xenctrl.h --- a/tools/libxc/xenctrl.h Wed Aug 30 13:51:12 2006 +0100 +++ b/tools/libxc/xenctrl.h Wed Aug 30 15:11:20 2006 -0500 @@ -416,6 +416,10 @@ int xc_domain_memory_populate_physmap(in unsigned int address_bits, xen_pfn_t

Last year we had a discussion[1] about how the hypercall ABI unfortunately contains fields that change width between 32- and 64-bit builds. This is a huge problem as we come up on the python management stack for ppc64, since the distributions ship 32-bit python. A 32-bit python/libxc cannot currently manage a 64-bit hypervisor. I had a patch but was unable to test it, and some other things were

This topic came up months ago (under the thread "make hypercall_preempt_check() a little more sensitive"), but since then, due to a misunderstanding, PPC has been running with a hack instead of a solution. So anyways, I''ve been digging into this again. PowerPC domains are allowed to write to the "interrupts enabled" bit (called External Exceptions, or EE) in the

As mentioned earlier, PowerPC''s atomic ops operate on longs, and we have made our *_bit() prototypes use long* (instead of void*) to warn us of problems at compile time. Here''s one caller that was flagged: test_and_set_bit(nr, &softirq_pending(cpu)) Accordingly, we need __softirq_pending to be long, not int. PowerPC is currently using a few files unmodified from the x86

I cleaned up the top level makefile in the tools directory. No major changes. Except I have it so that ioemmu is compiled only with x86_32. Signed-off-by: Jerone Young <jyoung5@us.ibm.com> --- tools/Makefile.orig 2005-03-17 21:03:44.000000000 -0600 +++ tools/Makefile 2005-03-22 15:05:20.000000000 -0600 @@ -1,37 +1,33 @@ +XEN_ROOT = ../ +include $(XEN_ROOT)/tools/Rules.mk -all: -

Hi, it looks like a few users of calloc had their arguments backwards. I checked the other users and they seem fine. Since one of those is in ioemu code, does that mean we (I?) will be submitting that bug to qemu upstream? -- Hollis Blanchard IBM Linux Technology Center Fix swapped calloc() arguments. Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com> diff -r c4eead8a925b

This allows tools in the python directory to properly link to libxenstore.a on x86-64. --- tools/xenstore/Makefile.orig 2005-06-09 12:56:34.000000000 -0500 +++ tools/xenstore/Makefile 2005-06-09 13:48:06.000000000 -0500 @@ -20,6 +20,9 @@ BASECFLAGS+= -I. CFLAGS+=$(BASECFLAGS) +ifeq ($(XEN_TARGET_ARCH),x86_64) +CFLAGS += -fPIC +endif LDFLAGS=$(PROFILE) -L$(XEN_LIBXC)

changeset: 30726:2a6fda4e7dde1a0a5d29a62303e85bcea868eb47 tag: tip user: Jimi Xenidis <jimix@watson.ibm.com> date: Thu Jul 13 11:51:38 2006 -0400 files: drivers/xen/privcmd/Makefile drivers/xen/privcmd/compat_privcmd.c fs/compat_ioctl.c include/xen/public/privcmd.h description: [ppc] ioctl32() compat plumbing for xen calls The following patch deals with xen

Hi All I want to use software counters available via "xenperf". However I get this error "Error getting number of perf counters: 38 (Function not implemented)". Am I missing something? Some configuration or option which needs to be turned on before I can use these counters. Any help would be much appreciated. I am using Xen 3.1.0 x86_64 -- Jyotirmaya Tripathi Virginia Tech

On Tuesday 02 August 2005 10:42, Xen patchbot -unstable wrote: > # HG changeset patch > # User smh22@firebug.cl.cam.ac.uk > # Node ID 59e76450e286240decceda23eca343ec4604124f > # Parent 48dea637aac96bcbabe788d036b52570520cc82e > Fixes. Sorry, but could we not make checkin comments like "Fixes."? It would just take a few more seconds to describe it in a complete

Hi Rusty, I had a look over your 011-paravirt-head.S.patch. I'm struggling to come up with a list of any benefits over having separate entrypoints for each hypervisor. Multiple entry pros: * allows maximum startup flexibility for any given hypervisor * for Xen at least, it's pretty simple * the "what hypervisor am I under" question is answered trivially cons: