[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] [PATCH v11 10/12] xen/pvh: Piggyback on PVHVM for grant driver.
On Wed, 18 Dec 2013, Konrad Rzeszutek Wilk wrote: > On Wed, Dec 18, 2013 at 06:46:56PM +0000, Stefano Stabellini wrote: > > On Tue, 17 Dec 2013, Konrad Rzeszutek Wilk wrote: > > > In PVH the shared grant frame is the PFN and not MFN, > > > hence its mapped via the same code path as HVM. > > > > > > The allocation of the grant frame is done differently - we > > > do not use the early platform-pci driver and have an > > > ioremap area - instead we use balloon memory and stitch > > > all of the non-contingous pages in a virtualized area. > > > > > > That means when we call the hypervisor to replace the GMFN > > > with a XENMAPSPACE_grant_table type, we need to lookup the > > > old PFN for every iteration instead of assuming a flat > > > contingous PFN allocation. > > > > > > Lastly, we only use v1 for grants. This is because PVHVM > > > is not able to use v2 due to no XENMEM_add_to_physmap > > > calls on the error status page (see commit > > > 69e8f430e243d657c2053f097efebc2e2cd559f0 > > > xen/granttable: Disable grant v2 for HVM domains.) > > > > > > Until that is implemented this workaround has to > > > be in place. > > > > > > Also per suggestions by Stefano utilize the PVHVM paths > > > as they share common functionality. > > > > > > Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx> > > > --- > > > drivers/xen/gntdev.c | 2 +- > > > drivers/xen/grant-table.c | 80 > > > ++++++++++++++++++++++++++++++++++++++++++---- > > > drivers/xen/platform-pci.c | 2 +- > > > include/xen/grant_table.h | 2 +- > > > 4 files changed, 76 insertions(+), 10 deletions(-) > > > > > > diff --git a/drivers/xen/gntdev.c b/drivers/xen/gntdev.c > > > index e41c79c..073b4a1 100644 > > > --- a/drivers/xen/gntdev.c > > > +++ b/drivers/xen/gntdev.c > > > @@ -846,7 +846,7 @@ static int __init gntdev_init(void) > > > if (!xen_domain()) > > > return -ENODEV; > > > > > > - use_ptemod = xen_pv_domain(); > > > + use_ptemod = !xen_feature(XENFEAT_auto_translated_physmap); > > > > > > err = misc_register(&gntdev_miscdev); > > > if (err != 0) { > > > diff --git a/drivers/xen/grant-table.c b/drivers/xen/grant-table.c > > > index aa846a4..c0ded9f 100644 > > > --- a/drivers/xen/grant-table.c > > > +++ b/drivers/xen/grant-table.c > > > @@ -47,6 +47,7 @@ > > > #include <xen/interface/xen.h> > > > #include <xen/page.h> > > > #include <xen/grant_table.h> > > > +#include <xen/balloon.h> > > > #include <xen/interface/memory.h> > > > #include <xen/hvc-console.h> > > > #include <xen/swiotlb-xen.h> > > > @@ -66,8 +67,8 @@ static unsigned int boot_max_nr_grant_frames; > > > static int gnttab_free_count; > > > static grant_ref_t gnttab_free_head; > > > static DEFINE_SPINLOCK(gnttab_list_lock); > > > -unsigned long xen_hvm_resume_frames; > > > -EXPORT_SYMBOL_GPL(xen_hvm_resume_frames); > > > +unsigned long xen_auto_xlat_grant_frames; > > > +EXPORT_SYMBOL_GPL(xen_auto_xlat_grant_frames); > > > > > > static union { > > > struct grant_entry_v1 *v1; > > > @@ -1060,7 +1061,7 @@ static int gnttab_map(unsigned int start_idx, > > > unsigned int end_idx) > > > unsigned int nr_gframes = end_idx + 1; > > > int rc; > > > > > > - if (xen_hvm_domain()) { > > > + if (xen_feature(XENFEAT_auto_translated_physmap)) { > > > struct xen_add_to_physmap xatp; > > > unsigned int i = end_idx; > > > rc = 0; > > > @@ -1069,10 +1070,24 @@ static int gnttab_map(unsigned int start_idx, > > > unsigned int end_idx) > > > * index, ensuring that the table will grow only once. > > > */ > > > do { > > > + unsigned long vaddr; > > > + unsigned int level; > > > + pte_t *pte; > > > + > > > xatp.domid = DOMID_SELF; > > > xatp.idx = i; > > > xatp.space = XENMAPSPACE_grant_table; > > > - xatp.gpfn = (xen_hvm_resume_frames >> PAGE_SHIFT) + i; > > > + > > > + /* > > > + * Don't assume the memory is contingous. Lookup each. > > > + */ > > > + vaddr = xen_auto_xlat_grant_frames + (i * PAGE_SIZE); > > > + if (xen_hvm_domain()) > > > + xatp.gpfn = vaddr >> PAGE_SHIFT; > > > + else { > > > + pte = lookup_address(vaddr, &level); > > > > lookup_address is x86 only. I added an empty implementation under > > arch/arm to be able to compile stuff under drivers/xen but I would like > > to get rid of it. > > If you can't find a way to replace lookup_address with something that > > works on arm and arm64 too, you could always turn > > xen_auto_xlat_grant_frames into an array and store the pfns there. > > Yeah, that would work nicely too. Thought we would have to store the > size somewhere. Maybe a struct then. > > struct xlat_grant_frames { > int nr; > xen_pfn_t *array; > struct page *page; > }; sounds good > > > + } > > > rc = HYPERVISOR_memory_op(XENMEM_add_to_physmap, &xatp); > > > if (rc != 0) { > > > pr_warn("grant table add_to_physmap failed, > > > err=%d\n", > > > @@ -1135,7 +1150,7 @@ static void gnttab_request_version(void) > > > int rc; > > > struct gnttab_set_version gsv; > > > > > > - if (xen_hvm_domain()) > > > + if (xen_hvm_domain() || xen_feature(XENFEAT_auto_translated_physmap)) > > > > just turn this to > > > > if (xen_feature(XENFEAT_auto_translated_physmap)) > > Hmm, I would rather actually keep that until I fix the underlaying issue > of why v2 can't be used. And then we can get rid of this workaround. Given that HVM guests have XENFEAT_auto_translated_physmap, the two are equivalent. > > > gsv.version = 1; > > > else > > > gsv.version = 2; > > > @@ -1161,6 +1176,46 @@ static void gnttab_request_version(void) > > > pr_info("Grant tables using version %d layout\n", grant_table_version); > > > } > > > > > > +static int xlated_setup_gnttab_pages(unsigned long nr_grant_frames, > > > + unsigned long max, void *addr) > > > +{ > > > + struct page **pages; > > > + unsigned long *pfns; > > > + int rc, i; > > > + > > > + pages = kcalloc(nr_grant_frames, sizeof(pages[0]), GFP_KERNEL); > > > + if (!pages) > > > + return -ENOMEM; > > > + > > > + pfns = kcalloc(nr_grant_frames, sizeof(pfns[0]), GFP_KERNEL); > > > + if (!pfns) { > > > + kfree(pages); > > > + return -ENOMEM; > > > + } > > > + rc = alloc_xenballooned_pages(nr_grant_frames, pages, 0 /* lowmem */); > > > + if (rc) { > > > + pr_warn("%s Couldn't balloon alloc %ld pfns rc:%d\n", __func__, > > > + nr_grant_frames, rc); > > > + kfree(pages); > > > + kfree(pfns); > > > + return rc; > > > + } > > > + for (i = 0; i < nr_grant_frames; i++) > > > + pfns[i] = page_to_pfn(pages[i]); > > > + > > > + rc = arch_gnttab_map_shared(pfns, nr_grant_frames, max, addr); > > > + > > > + kfree(pages); > > > + kfree(pfns); > > > + if (rc) { > > > + pr_warn("%s Couldn't map %ld pfns rc:%d\n", __func__, > > > + nr_grant_frames, rc); > > > + free_xenballooned_pages(nr_grant_frames, pages); > > > + return rc; > > > + } > > > + return rc; > > > +} > > > > Please move this function somewhere under arch/x86/xen and call it from > > xen_start_kernel, it doesn't belong to common code. > > <nods> > > Actually, looking at how the xen-platform-pci does it. I am wondering > if that can be just made in one function that will do the right > thing. > > Can the xen-platform-pci use balloon pages instead of the MMIO BARs? > Or would that be a waste of memory? I think they could be used but it would be a waste and they would need to be contiguous in gpfn space. _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxx http://lists.xen.org/xen-devel
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |