|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Re: [Xen-devel] [PATCH v2 5/6] ioreq-server: add support for multiple servers
> -----Original Message----- > From: Andrew Cooper [mailto:andrew.cooper3@xxxxxxxxxx] > Sent: 04 March 2014 12:06 > To: Paul Durrant > Cc: xen-devel@xxxxxxxxxxxxx > Subject: Re: [Xen-devel] [PATCH v2 5/6] ioreq-server: add support for > multiple servers > > On 04/03/14 11:40, Paul Durrant wrote: > > The legacy 'catch-all' server is always created with id 0. Secondary > > servers will have an id ranging from 1 to a limit set by the toolstack > > via the 'max_emulators' build info field. This defaults to 1 so ordinarily > > no extra special pages are reserved for secondary emulators. It may be > > increased using the secondary_device_emulators parameter in xl.cfg(5). > > There's no clear limit to apply to the number of emulators so I've not > > applied one. > > > > Because of the re-arrangement of the special pages in a previous patch we > > only need the addition of parameter HVM_PARAM_NR_IOREQ_SERVERS > to determine > > the layout of the shared pages for multiple emulators. Guests migrated in > > from hosts without this patch will be lacking the save record which stores > > the new parameter and so the guest is assumed to only have had a single > > emulator. > > > > Added some more emacs boilerplate to xenctrl.h and xenguest.h > > > > Signed-off-by: Paul Durrant <paul.durrant@xxxxxxxxxx> > > How does the build param interact with the hvmparam? It appears not to. > The build param is the value used to set the HVM param on domain create. (See setup_guest() just below where all the special PFNs are set). > On migrate, the receiving side will have to know, out-of-band, what to > set max_emulators to when building the domain. The setparam code needs > to validate the hvmparam against the build param and return > -EINVAL/-E2BIG in the case that the hvmparam is too large. > xc_domain_restore() needs to detect this and abort the migration if the > guest can't be restored with the expected number of emulators. > I don't think we need any of this. The HVM param is in the save record and so will be preserved. Migration in from a prior version of Xen will initialize the value to 1, which is correct for that case. The only need for max_emulators is to reserve the correct number of special pages when a new domain is built, so this is irrelevant for a domain restore; the only important thing is that the maximum number of emulators should never exceed the value of max_emulators when the domain was originally built, and the preservation of the HVM param ensures this. Paul > ~Andrew > > > --- > > docs/man/xl.cfg.pod.5 | 7 + > > tools/libxc/xc_domain.c | 175 +++++++ > > tools/libxc/xc_domain_restore.c | 20 + > > tools/libxc/xc_domain_save.c | 12 + > > tools/libxc/xc_hvm_build_x86.c | 24 +- > > tools/libxc/xenctrl.h | 51 ++ > > tools/libxc/xenguest.h | 12 + > > tools/libxc/xg_save_restore.h | 1 + > > tools/libxl/libxl.h | 8 + > > tools/libxl/libxl_create.c | 3 + > > tools/libxl/libxl_dom.c | 1 + > > tools/libxl/libxl_types.idl | 1 + > > tools/libxl/xl_cmdimpl.c | 3 + > > xen/arch/x86/hvm/hvm.c | 951 > +++++++++++++++++++++++++++++++++++--- > > xen/arch/x86/hvm/io.c | 2 +- > > xen/include/asm-x86/hvm/domain.h | 23 +- > > xen/include/asm-x86/hvm/hvm.h | 1 + > > xen/include/public/hvm/hvm_op.h | 70 +++ > > xen/include/public/hvm/ioreq.h | 1 + > > xen/include/public/hvm/params.h | 4 +- > > 20 files changed, 1300 insertions(+), 70 deletions(-) > > > > diff --git a/docs/man/xl.cfg.pod.5 b/docs/man/xl.cfg.pod.5 > > index e15a49f..0226c55 100644 > > --- a/docs/man/xl.cfg.pod.5 > > +++ b/docs/man/xl.cfg.pod.5 > > @@ -1281,6 +1281,13 @@ specified, enabling the use of XenServer PV > drivers in the guest. > > This parameter only takes effect when device_model_version=qemu-xen. > > See F<docs/misc/pci-device-reservations.txt> for more information. > > > > +=item B<secondary_device_emulators=NUMBER> > > + > > +If a number of secondary device emulators (i.e. in addition to > > +qemu-xen or qemu-xen-traditional) are to be invoked to support the > > +guest then this parameter can be set with the count of how many are > > +to be used. The default value is zero. > > + > > =back > > > > =head2 Device-Model Options > > diff --git a/tools/libxc/xc_domain.c b/tools/libxc/xc_domain.c > > index 369c3f3..dfa905b 100644 > > --- a/tools/libxc/xc_domain.c > > +++ b/tools/libxc/xc_domain.c > > @@ -1284,6 +1284,181 @@ int xc_get_hvm_param(xc_interface *handle, > domid_t dom, int param, unsigned long > > return rc; > > } > > > > +int xc_hvm_create_ioreq_server(xc_interface *xch, > > + domid_t domid, > > + ioservid_t *id) > > +{ > > + DECLARE_HYPERCALL; > > + DECLARE_HYPERCALL_BUFFER(xen_hvm_create_ioreq_server_t, arg); > > + int rc; > > + > > + arg = xc_hypercall_buffer_alloc(xch, arg, sizeof(*arg)); > > + if ( arg == NULL ) > > + return -1; > > + > > + hypercall.op = __HYPERVISOR_hvm_op; > > + hypercall.arg[0] = HVMOP_create_ioreq_server; > > + hypercall.arg[1] = HYPERCALL_BUFFER_AS_ARG(arg); > > + arg->domid = domid; > > + rc = do_xen_hypercall(xch, &hypercall); > > + *id = arg->id; > > + xc_hypercall_buffer_free(xch, arg); > > + return rc; > > +} > > + > > +int xc_hvm_get_ioreq_server_info(xc_interface *xch, > > + domid_t domid, > > + ioservid_t id, > > + xen_pfn_t *pfn, > > + xen_pfn_t *buf_pfn, > > + evtchn_port_t *buf_port) > > +{ > > + DECLARE_HYPERCALL; > > + DECLARE_HYPERCALL_BUFFER(xen_hvm_get_ioreq_server_info_t, > arg); > > + int rc; > > + > > + arg = xc_hypercall_buffer_alloc(xch, arg, sizeof(*arg)); > > + if ( arg == NULL ) > > + return -1; > > + > > + hypercall.op = __HYPERVISOR_hvm_op; > > + hypercall.arg[0] = HVMOP_get_ioreq_server_info; > > + hypercall.arg[1] = HYPERCALL_BUFFER_AS_ARG(arg); > > + arg->domid = domid; > > + arg->id = id; > > + rc = do_xen_hypercall(xch, &hypercall); > > + if ( rc != 0 ) > > + goto done; > > + > > + if ( pfn ) > > + *pfn = arg->pfn; > > + > > + if ( buf_pfn ) > > + *buf_pfn = arg->buf_pfn; > > + > > + if ( buf_port ) > > + *buf_port = arg->buf_port; > > + > > +done: > > + xc_hypercall_buffer_free(xch, arg); > > + return rc; > > +} > > + > > +int xc_hvm_map_io_range_to_ioreq_server(xc_interface *xch, domid_t > domid, > > + ioservid_t id, int is_mmio, > > + uint64_t start, uint64_t end) > > +{ > > + DECLARE_HYPERCALL; > > + > DECLARE_HYPERCALL_BUFFER(xen_hvm_map_io_range_to_ioreq_server_t, > arg); > > + int rc; > > + > > + arg = xc_hypercall_buffer_alloc(xch, arg, sizeof(*arg)); > > + if ( arg == NULL ) > > + return -1; > > + > > + hypercall.op = __HYPERVISOR_hvm_op; > > + hypercall.arg[0] = HVMOP_map_io_range_to_ioreq_server; > > + hypercall.arg[1] = HYPERCALL_BUFFER_AS_ARG(arg); > > + arg->domid = domid; > > + arg->id = id; > > + arg->is_mmio = is_mmio; > > + arg->start = start; > > + arg->end = end; > > + rc = do_xen_hypercall(xch, &hypercall); > > + xc_hypercall_buffer_free(xch, arg); > > + return rc; > > +} > > + > > +int xc_hvm_unmap_io_range_from_ioreq_server(xc_interface *xch, > domid_t domid, > > + ioservid_t id, int is_mmio, > > + uint64_t start) > > +{ > > + DECLARE_HYPERCALL; > > + > DECLARE_HYPERCALL_BUFFER(xen_hvm_unmap_io_range_from_ioreq_ser > ver_t, arg); > > + int rc; > > + > > + arg = xc_hypercall_buffer_alloc(xch, arg, sizeof(*arg)); > > + if ( arg == NULL ) > > + return -1; > > + > > + hypercall.op = __HYPERVISOR_hvm_op; > > + hypercall.arg[0] = HVMOP_unmap_io_range_from_ioreq_server; > > + hypercall.arg[1] = HYPERCALL_BUFFER_AS_ARG(arg); > > + arg->domid = domid; > > + arg->id = id; > > + arg->is_mmio = is_mmio; > > + arg->start = start; > > + rc = do_xen_hypercall(xch, &hypercall); > > + xc_hypercall_buffer_free(xch, arg); > > + return rc; > > +} > > + > > +int xc_hvm_map_pcidev_to_ioreq_server(xc_interface *xch, domid_t > domid, > > + ioservid_t id, uint16_t bdf) > > +{ > > + DECLARE_HYPERCALL; > > + > DECLARE_HYPERCALL_BUFFER(xen_hvm_map_pcidev_to_ioreq_server_t, > arg); > > + int rc; > > + > > + arg = xc_hypercall_buffer_alloc(xch, arg, sizeof(*arg)); > > + if ( arg == NULL ) > > + return -1; > > + > > + hypercall.op = __HYPERVISOR_hvm_op; > > + hypercall.arg[0] = HVMOP_map_pcidev_to_ioreq_server; > > + hypercall.arg[1] = HYPERCALL_BUFFER_AS_ARG(arg); > > + arg->domid = domid; > > + arg->id = id; > > + arg->bdf = bdf; > > + rc = do_xen_hypercall(xch, &hypercall); > > + xc_hypercall_buffer_free(xch, arg); > > + return rc; > > +} > > + > > +int xc_hvm_unmap_pcidev_from_ioreq_server(xc_interface *xch, > domid_t domid, > > + ioservid_t id, uint16_t bdf) > > +{ > > + DECLARE_HYPERCALL; > > + > DECLARE_HYPERCALL_BUFFER(xen_hvm_unmap_pcidev_from_ioreq_serve > r_t, arg); > > + int rc; > > + > > + arg = xc_hypercall_buffer_alloc(xch, arg, sizeof(*arg)); > > + if ( arg == NULL ) > > + return -1; > > + > > + hypercall.op = __HYPERVISOR_hvm_op; > > + hypercall.arg[0] = HVMOP_unmap_pcidev_from_ioreq_server; > > + hypercall.arg[1] = HYPERCALL_BUFFER_AS_ARG(arg); > > + arg->domid = domid; > > + arg->id = id; > > + arg->bdf = bdf; > > + rc = do_xen_hypercall(xch, &hypercall); > > + xc_hypercall_buffer_free(xch, arg); > > + return rc; > > +} > > + > > +int xc_hvm_destroy_ioreq_server(xc_interface *xch, > > + domid_t domid, > > + ioservid_t id) > > +{ > > + DECLARE_HYPERCALL; > > + DECLARE_HYPERCALL_BUFFER(xen_hvm_destroy_ioreq_server_t, arg); > > + int rc; > > + > > + arg = xc_hypercall_buffer_alloc(xch, arg, sizeof(*arg)); > > + if ( arg == NULL ) > > + return -1; > > + > > + hypercall.op = __HYPERVISOR_hvm_op; > > + hypercall.arg[0] = HVMOP_destroy_ioreq_server; > > + hypercall.arg[1] = HYPERCALL_BUFFER_AS_ARG(arg); > > + arg->domid = domid; > > + arg->id = id; > > + rc = do_xen_hypercall(xch, &hypercall); > > + xc_hypercall_buffer_free(xch, arg); > > + return rc; > > +} > > + > > int xc_domain_setdebugging(xc_interface *xch, > > uint32_t domid, > > unsigned int enable) > > diff --git a/tools/libxc/xc_domain_restore.c > b/tools/libxc/xc_domain_restore.c > > index 1f6ce50..3116653 100644 > > --- a/tools/libxc/xc_domain_restore.c > > +++ b/tools/libxc/xc_domain_restore.c > > @@ -746,6 +746,7 @@ typedef struct { > > uint64_t acpi_ioport_location; > > uint64_t viridian; > > uint64_t vm_generationid_addr; > > + uint64_t nr_ioreq_servers; > > > > struct toolstack_data_t tdata; > > } pagebuf_t; > > @@ -996,6 +997,16 @@ static int pagebuf_get_one(xc_interface *xch, > struct restore_ctx *ctx, > > DPRINTF("read generation id buffer address"); > > return pagebuf_get_one(xch, ctx, buf, fd, dom); > > > > + case XC_SAVE_ID_HVM_NR_IOREQ_SERVERS: > > + /* Skip padding 4 bytes then read the acpi ioport location. */ > > + if ( RDEXACT(fd, &buf->nr_ioreq_servers, sizeof(uint32_t)) || > > + RDEXACT(fd, &buf->nr_ioreq_servers, sizeof(uint64_t)) ) > > + { > > + PERROR("error reading the number of IOREQ servers"); > > + return -1; > > + } > > + return pagebuf_get_one(xch, ctx, buf, fd, dom); > > + > > default: > > if ( (count > MAX_BATCH_SIZE) || (count < 0) ) { > > ERROR("Max batch size exceeded (%d). Giving up.", count); > > @@ -1755,6 +1766,15 @@ int xc_domain_restore(xc_interface *xch, int > io_fd, uint32_t dom, > > if (pagebuf.viridian != 0) > > xc_set_hvm_param(xch, dom, HVM_PARAM_VIRIDIAN, 1); > > > > + if ( hvm ) { > > + int nr_ioreq_servers = pagebuf.nr_ioreq_servers; > > + > > + if ( nr_ioreq_servers == 0 ) > > + nr_ioreq_servers = 1; > > + > > + xc_set_hvm_param(xch, dom, HVM_PARAM_NR_IOREQ_SERVERS, > nr_ioreq_servers); > > + } > > + > > if (pagebuf.acpi_ioport_location == 1) { > > DBGPRINTF("Use new firmware ioport from the checkpoint\n"); > > xc_set_hvm_param(xch, dom, > HVM_PARAM_ACPI_IOPORTS_LOCATION, 1); > > diff --git a/tools/libxc/xc_domain_save.c b/tools/libxc/xc_domain_save.c > > index 42c4752..3293e29 100644 > > --- a/tools/libxc/xc_domain_save.c > > +++ b/tools/libxc/xc_domain_save.c > > @@ -1731,6 +1731,18 @@ int xc_domain_save(xc_interface *xch, int io_fd, > uint32_t dom, uint32_t max_iter > > PERROR("Error when writing the viridian flag"); > > goto out; > > } > > + > > + chunk.id = XC_SAVE_ID_HVM_NR_IOREQ_SERVERS; > > + chunk.data = 0; > > + xc_get_hvm_param(xch, dom, HVM_PARAM_NR_IOREQ_SERVERS, > > + (unsigned long *)&chunk.data); > > + > > + if ( (chunk.data != 0) && > > + wrexact(io_fd, &chunk, sizeof(chunk)) ) > > + { > > + PERROR("Error when writing the number of IOREQ servers"); > > + goto out; > > + } > > } > > > > if ( callbacks != NULL && callbacks->toolstack_save != NULL ) > > diff --git a/tools/libxc/xc_hvm_build_x86.c > b/tools/libxc/xc_hvm_build_x86.c > > index b65e702..6d6328a 100644 > > --- a/tools/libxc/xc_hvm_build_x86.c > > +++ b/tools/libxc/xc_hvm_build_x86.c > > @@ -45,7 +45,7 @@ > > #define SPECIALPAGE_IDENT_PT 4 > > #define SPECIALPAGE_CONSOLE 5 > > #define SPECIALPAGE_IOREQ 6 > > -#define NR_SPECIAL_PAGES SPECIALPAGE_IOREQ + 2 /* ioreq server > needs 2 pages */ > > +#define NR_SPECIAL_PAGES(n) SPECIALPAGE_IOREQ + (2 * n) /* ioreq > server needs 2 pages */ > > #define special_pfn(x) (0xff000u - 1 - (x)) > > > > #define VGA_HOLE_SIZE (0x20) > > @@ -85,7 +85,8 @@ static int modules_init(struct xc_hvm_build_args > *args, > > } > > > > static void build_hvm_info(void *hvm_info_page, uint64_t mem_size, > > - uint64_t mmio_start, uint64_t mmio_size) > > + uint64_t mmio_start, uint64_t mmio_size, > > + int max_emulators) > > { > > struct hvm_info_table *hvm_info = (struct hvm_info_table *) > > (((unsigned char *)hvm_info_page) + HVM_INFO_OFFSET); > > @@ -113,7 +114,7 @@ static void build_hvm_info(void *hvm_info_page, > uint64_t mem_size, > > /* Memory parameters. */ > > hvm_info->low_mem_pgend = lowmem_end >> PAGE_SHIFT; > > hvm_info->high_mem_pgend = highmem_end >> PAGE_SHIFT; > > - hvm_info->reserved_mem_pgstart = special_pfn(0) - > NR_SPECIAL_PAGES; > > + hvm_info->reserved_mem_pgstart = special_pfn(0) - > NR_SPECIAL_PAGES(max_emulators); > > > > /* Finish with the checksum. */ > > for ( i = 0, sum = 0; i < hvm_info->length; i++ ) > > @@ -256,6 +257,10 @@ static int setup_guest(xc_interface *xch, > > stat_1gb_pages = 0; > > int pod_mode = 0; > > int claim_enabled = args->claim_enabled; > > + int max_emulators = args->max_emulators; > > + > > + if ( max_emulators < 1 ) > > + goto error_out; > > > > if ( nr_pages > target_pages ) > > pod_mode = XENMEMF_populate_on_demand; > > @@ -468,12 +473,13 @@ static int setup_guest(xc_interface *xch, > > xch, dom, PAGE_SIZE, PROT_READ | PROT_WRITE, > > HVM_INFO_PFN)) == NULL ) > > goto error_out; > > - build_hvm_info(hvm_info_page, v_end, mmio_start, mmio_size); > > + build_hvm_info(hvm_info_page, v_end, mmio_start, mmio_size, > > + max_emulators); > > munmap(hvm_info_page, PAGE_SIZE); > > > > /* Allocate and clear special pages. */ > > > > - DPRINTF("%d SPECIAL PAGES:\n", NR_SPECIAL_PAGES); > > + DPRINTF("%d SPECIAL PAGES:\n", > NR_SPECIAL_PAGES(max_emulators)); > > DPRINTF(" PAGING: %"PRI_xen_pfn"\n", > > (xen_pfn_t)special_pfn(SPECIALPAGE_PAGING)); > > DPRINTF(" ACCESS: %"PRI_xen_pfn"\n", > > @@ -486,10 +492,10 @@ static int setup_guest(xc_interface *xch, > > (xen_pfn_t)special_pfn(SPECIALPAGE_IDENT_PT)); > > DPRINTF(" CONSOLE: %"PRI_xen_pfn"\n", > > (xen_pfn_t)special_pfn(SPECIALPAGE_CONSOLE)); > > - DPRINTF(" IOREQ: %"PRI_xen_pfn"\n", > > + DPRINTF(" IOREQ(%02d): %"PRI_xen_pfn"\n", max_emulators * 2, > > (xen_pfn_t)special_pfn(SPECIALPAGE_IOREQ)); > > > > - for ( i = 0; i < NR_SPECIAL_PAGES; i++ ) > > + for ( i = 0; i < NR_SPECIAL_PAGES(max_emulators); i++ ) > > { > > xen_pfn_t pfn = special_pfn(i); > > rc = xc_domain_populate_physmap_exact(xch, dom, 1, 0, 0, &pfn); > > @@ -515,7 +521,9 @@ static int setup_guest(xc_interface *xch, > > xc_set_hvm_param(xch, dom, HVM_PARAM_IOREQ_PFN, > > special_pfn(SPECIALPAGE_IOREQ)); > > xc_set_hvm_param(xch, dom, HVM_PARAM_BUFIOREQ_PFN, > > - special_pfn(SPECIALPAGE_IOREQ) - 1); > > + special_pfn(SPECIALPAGE_IOREQ) - max_emulators); > > + xc_set_hvm_param(xch, dom, HVM_PARAM_NR_IOREQ_SERVERS, > > + max_emulators); > > > > /* > > * Identity-map page table is required for running with CR0.PG=0 when > > diff --git a/tools/libxc/xenctrl.h b/tools/libxc/xenctrl.h > > index 13f816b..84cab13 100644 > > --- a/tools/libxc/xenctrl.h > > +++ b/tools/libxc/xenctrl.h > > @@ -1801,6 +1801,47 @@ void xc_clear_last_error(xc_interface *xch); > > int xc_set_hvm_param(xc_interface *handle, domid_t dom, int param, > unsigned long value); > > int xc_get_hvm_param(xc_interface *handle, domid_t dom, int param, > unsigned long *value); > > > > +/* > > + * IOREQ server API > > + */ > > +int xc_hvm_create_ioreq_server(xc_interface *xch, > > + domid_t domid, > > + ioservid_t *id); > > + > > +int xc_hvm_get_ioreq_server_info(xc_interface *xch, > > + domid_t domid, > > + ioservid_t id, > > + xen_pfn_t *pfn, > > + xen_pfn_t *buf_pfn, > > + evtchn_port_t *buf_port); > > + > > +int xc_hvm_map_io_range_to_ioreq_server(xc_interface *xch, > > + domid_t domid, > > + ioservid_t id, > > + int is_mmio, > > + uint64_t start, > > + uint64_t end); > > + > > +int xc_hvm_unmap_io_range_from_ioreq_server(xc_interface *xch, > > + domid_t domid, > > + ioservid_t id, > > + int is_mmio, > > + uint64_t start); > > + > > +int xc_hvm_map_pcidev_to_ioreq_server(xc_interface *xch, > > + domid_t domid, > > + ioservid_t id, > > + uint16_t bdf); > > + > > +int xc_hvm_unmap_pcidev_from_ioreq_server(xc_interface *xch, > > + domid_t domid, > > + ioservid_t id, > > + uint16_t bdf); > > + > > +int xc_hvm_destroy_ioreq_server(xc_interface *xch, > > + domid_t domid, > > + ioservid_t id); > > + > > /* HVM guest pass-through */ > > int xc_assign_device(xc_interface *xch, > > uint32_t domid, > > @@ -2428,3 +2469,13 @@ int xc_kexec_load(xc_interface *xch, uint8_t > type, uint16_t arch, > > int xc_kexec_unload(xc_interface *xch, int type); > > > > #endif /* XENCTRL_H */ > > + > > +/* > > + * Local variables: > > + * mode: C > > + * c-file-style: "BSD" > > + * c-basic-offset: 4 > > + * tab-width: 4 > > + * indent-tabs-mode: nil > > + * End: > > + */ > > diff --git a/tools/libxc/xenguest.h b/tools/libxc/xenguest.h > > index a0e30e1..1300933 100644 > > --- a/tools/libxc/xenguest.h > > +++ b/tools/libxc/xenguest.h > > @@ -234,6 +234,8 @@ struct xc_hvm_build_args { > > struct xc_hvm_firmware_module smbios_module; > > /* Whether to use claim hypercall (1 - enable, 0 - disable). */ > > int claim_enabled; > > + /* Maximum number of emulators for VM */ > > + int max_emulators; > > }; > > > > /** > > @@ -306,3 +308,13 @@ xen_pfn_t *xc_map_m2p(xc_interface *xch, > > int prot, > > unsigned long *mfn0); > > #endif /* XENGUEST_H */ > > + > > +/* > > + * Local variables: > > + * mode: C > > + * c-file-style: "BSD" > > + * c-basic-offset: 4 > > + * tab-width: 4 > > + * indent-tabs-mode: nil > > + * End: > > + */ > > diff --git a/tools/libxc/xg_save_restore.h b/tools/libxc/xg_save_restore.h > > index f859621..5170b7f 100644 > > --- a/tools/libxc/xg_save_restore.h > > +++ b/tools/libxc/xg_save_restore.h > > @@ -259,6 +259,7 @@ > > #define XC_SAVE_ID_HVM_ACCESS_RING_PFN -16 > > #define XC_SAVE_ID_HVM_SHARING_RING_PFN -17 > > #define XC_SAVE_ID_TOOLSTACK -18 /* Optional toolstack specific > info */ > > +#define XC_SAVE_ID_HVM_NR_IOREQ_SERVERS -19 > > > > /* > > ** We process save/restore/migrate in batches of pages; the below > > diff --git a/tools/libxl/libxl.h b/tools/libxl/libxl.h > > index 06bbca6..5a70b76 100644 > > --- a/tools/libxl/libxl.h > > +++ b/tools/libxl/libxl.h > > @@ -95,6 +95,14 @@ > > #define LIBXL_HAVE_BUILDINFO_EVENT_CHANNELS 1 > > > > /* > > + * LIBXL_HAVE_BUILDINFO_HVM_MAX_EMULATORS indicates that the > > + * max_emulators field is present in the hvm sections of > > + * libxl_domain_build_info. This field can be used to reserve > > + * extra special pages for secondary device emulators. > > + */ > > +#define LIBXL_HAVE_BUILDINFO_HVM_MAX_EMULATORS 1 > > + > > +/* > > * libxl ABI compatibility > > * > > * The only guarantee which libxl makes regarding ABI compatibility > > diff --git a/tools/libxl/libxl_create.c b/tools/libxl/libxl_create.c > > index a604cd8..cce93d9 100644 > > --- a/tools/libxl/libxl_create.c > > +++ b/tools/libxl/libxl_create.c > > @@ -330,6 +330,9 @@ int libxl__domain_build_info_setdefault(libxl__gc > *gc, > > > > libxl_defbool_setdefault(&b_info->u.hvm.gfx_passthru, false); > > > > + if (b_info->u.hvm.max_emulators < 1) > > + b_info->u.hvm.max_emulators = 1; > > + > > break; > > case LIBXL_DOMAIN_TYPE_PV: > > libxl_defbool_setdefault(&b_info->u.pv.e820_host, false); > > diff --git a/tools/libxl/libxl_dom.c b/tools/libxl/libxl_dom.c > > index 55f74b2..9de06f9 100644 > > --- a/tools/libxl/libxl_dom.c > > +++ b/tools/libxl/libxl_dom.c > > @@ -637,6 +637,7 @@ int libxl__build_hvm(libxl__gc *gc, uint32_t domid, > > args.mem_size = (uint64_t)(info->max_memkb - info->video_memkb) > << 10; > > args.mem_target = (uint64_t)(info->target_memkb - info- > >video_memkb) << 10; > > args.claim_enabled = libxl_defbool_val(info->claim_mode); > > + args.max_emulators = info->u.hvm.max_emulators; > > if (libxl__domain_firmware(gc, info, &args)) { > > LOG(ERROR, "initializing domain firmware failed"); > > goto out; > > diff --git a/tools/libxl/libxl_types.idl b/tools/libxl/libxl_types.idl > > index 649ce50..b707159 100644 > > --- a/tools/libxl/libxl_types.idl > > +++ b/tools/libxl/libxl_types.idl > > @@ -372,6 +372,7 @@ libxl_domain_build_info = > Struct("domain_build_info",[ > > ("xen_platform_pci", libxl_defbool), > > ("usbdevice_list", > > libxl_string_list), > > ("vendor_device", > > libxl_vendor_device), > > + ("max_emulators", integer), > > ])), > > ("pv", Struct(None, [("kernel", string), > > ("slack_memkb", MemKB), > > diff --git a/tools/libxl/xl_cmdimpl.c b/tools/libxl/xl_cmdimpl.c > > index 4fc46eb..cf9b67d 100644 > > --- a/tools/libxl/xl_cmdimpl.c > > +++ b/tools/libxl/xl_cmdimpl.c > > @@ -1750,6 +1750,9 @@ skip_vfb: > > > > b_info->u.hvm.vendor_device = d; > > } > > + > > + if (!xlu_cfg_get_long (config, "secondary_device_emulators", &l, > > 0)) > > + b_info->u.hvm.max_emulators = l + 1; > > } > > > > xlu_cfg_destroy(config); > > diff --git a/xen/arch/x86/hvm/hvm.c b/xen/arch/x86/hvm/hvm.c > > index fb2dd73..e8b73fa 100644 > > --- a/xen/arch/x86/hvm/hvm.c > > +++ b/xen/arch/x86/hvm/hvm.c > > @@ -357,14 +357,21 @@ static ioreq_t *get_ioreq(struct > hvm_ioreq_server *s, int id) > > bool_t hvm_io_pending(struct vcpu *v) > > { > > struct domain *d = v->domain; > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > - ioreq_t *p; > > + struct list_head *entry; > > > > - if ( !s ) > > - return 0; > > + list_for_each ( entry, &d->arch.hvm_domain.ioreq_server_list ) > > + { > > + struct hvm_ioreq_server *s = list_entry(entry, > > + struct hvm_ioreq_server, > > + list_entry); > > + ioreq_t *p = get_ioreq(s, v->vcpu_id); > > > > - p = get_ioreq(s, v->vcpu_id); > > - return ( p->state != STATE_IOREQ_NONE ); > > + p = get_ioreq(s, v->vcpu_id); > > + if ( p->state != STATE_IOREQ_NONE ) > > + return 1; > > + } > > + > > + return 0; > > } > > > > static void hvm_wait_on_io(struct domain *d, ioreq_t *p) > > @@ -394,18 +401,20 @@ static void hvm_wait_on_io(struct domain *d, > ioreq_t *p) > > void hvm_do_resume(struct vcpu *v) > > { > > struct domain *d = v->domain; > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > + struct list_head *entry; > > > > check_wakeup_from_wait(); > > > > if ( is_hvm_vcpu(v) ) > > pt_restore_timer(v); > > > > - if ( s ) > > + list_for_each ( entry, &d->arch.hvm_domain.ioreq_server_list ) > > { > > - ioreq_t *p = get_ioreq(s, v->vcpu_id); > > + struct hvm_ioreq_server *s = list_entry(entry, > > + struct hvm_ioreq_server, > > + list_entry); > > > > - hvm_wait_on_io(d, p); > > + hvm_wait_on_io(d, get_ioreq(s, v->vcpu_id)); > > } > > > > /* Inject pending hw/sw trap */ > > @@ -543,6 +552,83 @@ static int hvm_print_line( > > return X86EMUL_OKAY; > > } > > > > +static int hvm_access_cf8( > > + int dir, uint32_t port, uint32_t bytes, uint32_t *val) > > +{ > > + struct vcpu *curr = current; > > + struct hvm_domain *hd = &curr->domain->arch.hvm_domain; > > + int rc; > > + > > + BUG_ON(port < 0xcf8); > > + port -= 0xcf8; > > + > > + spin_lock(&hd->pci_lock); > > + > > + if ( dir == IOREQ_WRITE ) > > + { > > + switch ( bytes ) > > + { > > + case 4: > > + hd->pci_cf8 = *val; > > + break; > > + > > + case 2: > > + { > > + uint32_t mask = 0xffff << (port * 8); > > + uint32_t subval = *val << (port * 8); > > + > > + hd->pci_cf8 = (hd->pci_cf8 & ~mask) | > > + (subval & mask); > > + break; > > + } > > + > > + case 1: > > + { > > + uint32_t mask = 0xff << (port * 8); > > + uint32_t subval = *val << (port * 8); > > + > > + hd->pci_cf8 = (hd->pci_cf8 & ~mask) | > > + (subval & mask); > > + break; > > + } > > + > > + default: > > + break; > > + } > > + > > + /* We always need to fall through to the catch all emulator */ > > + rc = X86EMUL_UNHANDLEABLE; > > + } > > + else > > + { > > + switch ( bytes ) > > + { > > + case 4: > > + *val = hd->pci_cf8; > > + rc = X86EMUL_OKAY; > > + break; > > + > > + case 2: > > + *val = (hd->pci_cf8 >> (port * 8)) & 0xffff; > > + rc = X86EMUL_OKAY; > > + break; > > + > > + case 1: > > + *val = (hd->pci_cf8 >> (port * 8)) & 0xff; > > + rc = X86EMUL_OKAY; > > + break; > > + > > + default: > > + rc = X86EMUL_UNHANDLEABLE; > > + break; > > + } > > + } > > + > > + spin_unlock(&hd->pci_lock); > > + > > + return rc; > > +} > > + > > static int handle_pvh_io( > > int dir, uint32_t port, uint32_t bytes, uint32_t *val) > > { > > @@ -618,39 +704,53 @@ static void > hvm_ioreq_server_remove_vcpu(struct hvm_ioreq_server *s, struct vcpu > > } > > } > > > > -static int hvm_create_ioreq_server(struct domain *d, domid_t domid) > > +static int hvm_create_ioreq_server(struct domain *d, ioservid_t id, > domid_t domid) > > { > > struct hvm_ioreq_server *s; > > unsigned long pfn; > > struct vcpu *v; > > int i, rc; > > > > + if ( id >= d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS] ) > > + return -EINVAL; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > rc = -EEXIST; > > - if ( d->arch.hvm_domain.ioreq_server != NULL ) > > - goto fail_exist; > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry ) > > + { > > + if ( s->id == id ) > > + goto fail_exist; > > + } > > > > - gdprintk(XENLOG_INFO, "%s: %d\n", __func__, d->domain_id); > > + gdprintk(XENLOG_INFO, "%s: %d:%d\n", __func__, d->domain_id, id); > > > > rc = -ENOMEM; > > s = xzalloc(struct hvm_ioreq_server); > > if ( !s ) > > goto fail_alloc; > > > > + s->id = id; > > s->domain = d; > > s->domid = domid; > > + INIT_LIST_HEAD(&s->mmio_range_list); > > + INIT_LIST_HEAD(&s->portio_range_list); > > + INIT_LIST_HEAD(&s->pcidev_list); > > > > for ( i = 0; i < MAX_HVM_VCPUS; i++ ) > > s->ioreq_evtchn[i] = -1; > > s->buf_ioreq_evtchn = -1; > > > > /* Initialize shared pages */ > > - pfn = d->arch.hvm_domain.params[HVM_PARAM_IOREQ_PFN]; > > + pfn = d->arch.hvm_domain.params[HVM_PARAM_IOREQ_PFN] - s->id; > > > > hvm_init_ioreq_page(s, 0); > > if ( (rc = hvm_set_ioreq_page(s, 0, pfn)) < 0 ) > > goto fail_set_ioreq; > > > > - pfn = d->arch.hvm_domain.params[HVM_PARAM_BUFIOREQ_PFN]; > > + pfn = d->arch.hvm_domain.params[HVM_PARAM_BUFIOREQ_PFN] - s- > >id; > > > > hvm_init_ioreq_page(s, 1); > > if ( (rc = hvm_set_ioreq_page(s, 1, pfn)) < 0 ) > > @@ -664,10 +764,12 @@ static int hvm_create_ioreq_server(struct domain > *d, domid_t domid) > > goto fail_add_vcpu; > > } > > > > - d->arch.hvm_domain.ioreq_server = s; > > + list_add(&s->list_entry, > > + &d->arch.hvm_domain.ioreq_server_list); > > > > domain_unpause(d); > > > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > return 0; > > > > fail_add_vcpu: > > @@ -681,23 +783,33 @@ fail_set_ioreq: > > xfree(s); > > fail_alloc: > > fail_exist: > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > return rc; > > } > > > > -static void hvm_destroy_ioreq_server(struct domain *d) > > +static void hvm_destroy_ioreq_server(struct domain *d, ioservid_t id) > > { > > struct hvm_ioreq_server *s; > > struct vcpu *v; > > > > - gdprintk(XENLOG_INFO, "%s: %d\n", __func__, d->domain_id); > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > > > - s = d->arch.hvm_domain.ioreq_server; > > - if ( !s ) > > - return; > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry) > > + { > > + if ( s->id == id ) > > + goto found; > > + } > > + > > + goto done; > > + > > +found: > > + gdprintk(XENLOG_INFO, "%s: %d:%d\n", __func__, d->domain_id, id); > > > > domain_pause(d); > > > > - d->arch.hvm_domain.ioreq_server = NULL; > > + list_del_init(&s->list_entry); > > > > for_each_vcpu ( d, v ) > > hvm_ioreq_server_remove_vcpu(s, v); > > @@ -708,31 +820,373 @@ static void hvm_destroy_ioreq_server(struct > domain *d) > > hvm_destroy_ioreq_page(s, 0); > > > > xfree(s); > > + > > +done: > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > } > > > > -static int hvm_get_ioreq_server_buf_port(struct domain *d, > evtchn_port_t *port) > > +static int hvm_get_ioreq_server_buf_port(struct domain *d, ioservid_t > id, > > + evtchn_port_t *port) > > { > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > + struct list_head *entry; > > + int rc; > > > > - if ( !s ) > > - return -ENOENT; > > + if ( id >= d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS] ) > > + return -EINVAL; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + rc = -ENOENT; > > + list_for_each ( entry, > > + &d->arch.hvm_domain.ioreq_server_list ) > > + { > > + struct hvm_ioreq_server *s = list_entry(entry, > > + struct hvm_ioreq_server, > > + list_entry); > > + > > + if ( s->id == id ) > > + { > > + *port = s->buf_ioreq_evtchn; > > + rc = 0; > > + break; > > + } > > + } > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + return rc; > > +} > > + > > +static int hvm_get_ioreq_server_pfn(struct domain *d, ioservid_t id, int > buf, > > + xen_pfn_t *pfn) > > +{ > > + struct list_head *entry; > > + int rc; > > + > > + if ( id >= d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS] ) > > + return -EINVAL; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + rc = -ENOENT; > > + list_for_each ( entry, > > + &d->arch.hvm_domain.ioreq_server_list ) > > + { > > + struct hvm_ioreq_server *s = list_entry(entry, > > + struct hvm_ioreq_server, > > + list_entry); > > + > > + if ( s->id == id ) > > + { > > + int i = ( buf ) ? HVM_PARAM_BUFIOREQ_PFN : > HVM_PARAM_IOREQ_PFN; > > + > > + *pfn = d->arch.hvm_domain.params[i] - s->id; > > + rc = 0; > > + break; > > + } > > + } > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + return rc; > > +} > > + > > +static int hvm_map_io_range_to_ioreq_server(struct domain *d, > ioservid_t id, > > + int is_mmio, uint64_t start, > > uint64_t end) > > +{ > > + struct hvm_ioreq_server *s; > > + struct hvm_io_range *x; > > + struct list_head *list; > > + int rc; > > + > > + if ( id >= d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS] ) > > + return -EINVAL; > > + > > + x = xmalloc(struct hvm_io_range); > > + if ( x == NULL ) > > + return -ENOMEM; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + rc = -ENOENT; > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry ) > > + { > > + if ( s->id == id ) > > + goto found; > > + } > > + > > + goto fail; > > + > > +found: > > + INIT_RCU_HEAD(&x->rcu); > > + x->start = start; > > + x->end = end; > > + > > + list = ( is_mmio ) ? &s->mmio_range_list : &s->portio_range_list; > > + list_add_rcu(&x->list_entry, list); > > + > > + gdprintk(XENLOG_DEBUG, "%d:%d: +%s %"PRIX64" - %"PRIX64"\n", > > + d->domain_id, > > + s->id, > > + ( is_mmio ) ? "MMIO" : "PORTIO", > > + x->start, > > + x->end); > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > > > - *port = s->buf_ioreq_evtchn; > > return 0; > > + > > +fail: > > + xfree(x); > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + return rc; > > } > > > > -static int hvm_get_ioreq_server_pfn(struct domain *d, int buf, xen_pfn_t > *pfn) > > +static void free_io_range(struct rcu_head *rcu) > > { > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > - int i; > > + struct hvm_io_range *x; > > > > - if ( !s ) > > - return -ENOENT; > > + x = container_of (rcu, struct hvm_io_range, rcu); > > + > > + xfree(x); > > +} > > + > > +static int hvm_unmap_io_range_from_ioreq_server(struct domain *d, > ioservid_t id, > > + int is_mmio, uint64_t > > start) > > +{ > > + struct hvm_ioreq_server *s; > > + struct list_head *list, *entry; > > + int rc; > > + > > + if ( id >= d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS] ) > > + return -EINVAL; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + rc = -ENOENT; > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry ) > > + { > > + if ( s->id == id ) > > + goto found; > > + } > > + > > + goto done; > > + > > +found: > > + list = ( is_mmio ) ? &s->mmio_range_list : &s->portio_range_list; > > + > > + list_for_each ( entry, > > + list ) > > + { > > + struct hvm_io_range *x = list_entry(entry, > > + struct hvm_io_range, > > + list_entry); > > + > > + if ( start == x->start ) > > + { > > + gdprintk(XENLOG_DEBUG, "%d:%d: -%s %"PRIX64" - %"PRIX64"\n", > > + d->domain_id, > > + s->id, > > + ( is_mmio ) ? "MMIO" : "PORTIO", > > + x->start, > > + x->end); > > + > > + list_del_rcu(&x->list_entry); > > + call_rcu(&x->rcu, free_io_range); > > > > - i = ( buf ) ? HVM_PARAM_BUFIOREQ_PFN : HVM_PARAM_IOREQ_PFN; > > - *pfn = d->arch.hvm_domain.params[i]; > > + rc = 0; > > + break; > > + } > > + } > > + > > +done: > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + return rc; > > +} > > + > > +static int hvm_map_pcidev_to_ioreq_server(struct domain *d, ioservid_t > id, > > + uint16_t bdf) > > +{ > > + struct hvm_ioreq_server *s; > > + struct hvm_pcidev *x; > > + int rc; > > + > > + if ( id >= d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS] ) > > + return -EINVAL; > > + > > + x = xmalloc(struct hvm_pcidev); > > + if ( x == NULL ) > > + return -ENOMEM; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + rc = -ENOENT; > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry ) > > + { > > + if ( s->id == id ) > > + goto found; > > + } > > + > > + goto fail; > > + > > +found: > > + INIT_RCU_HEAD(&x->rcu); > > + x->bdf = bdf; > > + > > + list_add_rcu(&x->list_entry, &s->pcidev_list); > > + > > + gdprintk(XENLOG_DEBUG, "%d:%d: +PCIDEV %04X\n", > > + d->domain_id, > > + s->id, > > + x->bdf); > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > > > return 0; > > + > > +fail: > > + xfree(x); > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + return rc; > > +} > > + > > +static void free_pcidev(struct rcu_head *rcu) > > +{ > > + struct hvm_pcidev *x; > > + > > + x = container_of (rcu, struct hvm_pcidev, rcu); > > + > > + xfree(x); > > +} > > + > > +static int hvm_unmap_pcidev_from_ioreq_server(struct domain *d, > ioservid_t id, > > + uint16_t bdf) > > +{ > > + struct hvm_ioreq_server *s; > > + struct list_head *entry; > > + int rc; > > + > > + if ( id >= d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS] ) > > + return -EINVAL; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + rc = -ENOENT; > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry ) > > + { > > + if ( s->id == id ) > > + goto found; > > + } > > + > > + goto done; > > + > > +found: > > + list_for_each ( entry, > > + &s->pcidev_list ) > > + { > > + struct hvm_pcidev *x = list_entry(entry, > > + struct hvm_pcidev, > > + list_entry); > > + > > + if ( bdf == x->bdf ) > > + { > > + gdprintk(XENLOG_DEBUG, "%d:%d: -PCIDEV %04X\n", > > + d->domain_id, > > + s->id, > > + x->bdf); > > + > > + list_del_rcu(&x->list_entry); > > + call_rcu(&x->rcu, free_pcidev); > > + > > + rc = 0; > > + break; > > + } > > + } > > + > > +done: > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + return rc; > > +} > > + > > +static int hvm_all_ioreq_servers_add_vcpu(struct domain *d, struct vcpu > *v) > > +{ > > + struct list_head *entry; > > + int rc; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + list_for_each ( entry, > > + &d->arch.hvm_domain.ioreq_server_list ) > > + { > > + struct hvm_ioreq_server *s = list_entry(entry, > > + struct hvm_ioreq_server, > > + list_entry); > > + > > + if ( (rc = hvm_ioreq_server_add_vcpu(s, v)) < 0 ) > > + goto fail; > > + } > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + return 0; > > + > > +fail: > > + list_for_each ( entry, > > + &d->arch.hvm_domain.ioreq_server_list ) > > + { > > + struct hvm_ioreq_server *s = list_entry(entry, > > + struct hvm_ioreq_server, > > + list_entry); > > + > > + hvm_ioreq_server_remove_vcpu(s, v); > > + } > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + return rc; > > +} > > + > > +static void hvm_all_ioreq_servers_remove_vcpu(struct domain *d, struct > vcpu *v) > > +{ > > + struct list_head *entry; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > + list_for_each ( entry, > > + &d->arch.hvm_domain.ioreq_server_list ) > > + { > > + struct hvm_ioreq_server *s = list_entry(entry, > > + struct hvm_ioreq_server, > > + list_entry); > > + > > + hvm_ioreq_server_remove_vcpu(s, v); > > + } > > + > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > +} > > + > > +static void hvm_destroy_all_ioreq_servers(struct domain *d) > > +{ > > + ioservid_t id; > > + > > + for ( id = 0; > > + id < d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS]; > > + id++ ) > > + hvm_destroy_ioreq_server(d, id); > > } > > > > static int hvm_replace_event_channel(struct vcpu *v, domid_t > remote_domid, > > @@ -750,18 +1204,31 @@ static int hvm_replace_event_channel(struct > vcpu *v, domid_t remote_domid, > > return 0; > > } > > > > -static int hvm_set_ioreq_server_domid(struct domain *d, domid_t > domid) > > +static int hvm_set_ioreq_server_domid(struct domain *d, ioservid_t id, > domid_t domid) > > { > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > + struct hvm_ioreq_server *s; > > struct vcpu *v; > > int rc = 0; > > > > + if ( id >= d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS] ) > > + return -EINVAL; > > + > > + spin_lock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > domain_pause(d); > > > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry ) > > + { > > + if ( s->id == id ) > > + goto found; > > + } > > + > > rc = -ENOENT; > > - if ( !s ) > > - goto done; > > + goto done; > > > > +found: > > rc = 0; > > if ( s->domid == domid ) > > goto done; > > @@ -787,6 +1254,8 @@ static int hvm_set_ioreq_server_domid(struct > domain *d, domid_t domid) > > done: > > domain_unpause(d); > > > > + spin_unlock(&d->arch.hvm_domain.ioreq_server_lock); > > + > > return rc; > > } > > > > @@ -817,6 +1286,9 @@ int hvm_domain_initialise(struct domain *d) > > > > } > > > > + spin_lock_init(&d->arch.hvm_domain.ioreq_server_lock); > > + INIT_LIST_HEAD(&d->arch.hvm_domain.ioreq_server_list); > > + spin_lock_init(&d->arch.hvm_domain.pci_lock); > > spin_lock_init(&d->arch.hvm_domain.irq_lock); > > spin_lock_init(&d->arch.hvm_domain.uc_lock); > > > > @@ -858,6 +1330,7 @@ int hvm_domain_initialise(struct domain *d) > > rtc_init(d); > > > > register_portio_handler(d, 0xe9, 1, hvm_print_line); > > + register_portio_handler(d, 0xcf8, 4, hvm_access_cf8); > > > > rc = hvm_funcs.domain_initialise(d); > > if ( rc != 0 ) > > @@ -888,7 +1361,7 @@ void hvm_domain_relinquish_resources(struct > domain *d) > > if ( hvm_funcs.nhvm_domain_relinquish_resources ) > > hvm_funcs.nhvm_domain_relinquish_resources(d); > > > > - hvm_destroy_ioreq_server(d); > > + hvm_destroy_all_ioreq_servers(d); > > > > msixtbl_pt_cleanup(d); > > > > @@ -1520,7 +1993,6 @@ int hvm_vcpu_initialise(struct vcpu *v) > > { > > int rc; > > struct domain *d = v->domain; > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > > > hvm_asid_flush_vcpu(v); > > > > @@ -1563,12 +2035,9 @@ int hvm_vcpu_initialise(struct vcpu *v) > > && (rc = nestedhvm_vcpu_initialise(v)) < 0 ) /* teardown: > nestedhvm_vcpu_destroy */ > > goto fail5; > > > > - if ( s ) > > - { > > - rc = hvm_ioreq_server_add_vcpu(s, v); > > - if ( rc < 0 ) > > - goto fail6; > > - } > > + rc = hvm_all_ioreq_servers_add_vcpu(d, v); > > + if ( rc < 0 ) > > + goto fail6; > > > > if ( v->vcpu_id == 0 ) > > { > > @@ -1604,10 +2073,8 @@ int hvm_vcpu_initialise(struct vcpu *v) > > void hvm_vcpu_destroy(struct vcpu *v) > > { > > struct domain *d = v->domain; > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > > > - if ( s ) > > - hvm_ioreq_server_remove_vcpu(s, v); > > + hvm_all_ioreq_servers_remove_vcpu(d, v); > > > > nestedhvm_vcpu_destroy(v); > > > > @@ -1646,11 +2113,112 @@ void hvm_vcpu_down(struct vcpu *v) > > } > > } > > > > +static DEFINE_RCU_READ_LOCK(ioreq_server_rcu_lock); > > + > > +static struct hvm_ioreq_server *hvm_select_ioreq_server(struct vcpu *v, > ioreq_t *p) > > +{ > > +#define BDF(cf8) (((cf8) & 0x00ffff00) >> 8) > > + > > + struct domain *d = v->domain; > > + struct hvm_ioreq_server *s; > > + uint8_t type; > > + uint64_t addr; > > + > > + if ( p->type == IOREQ_TYPE_PIO && > > + (p->addr & ~3) == 0xcfc ) > > + { > > + /* PCI config data cycle */ > > + type = IOREQ_TYPE_PCI_CONFIG; > > + > > + spin_lock(&d->arch.hvm_domain.pci_lock); > > + addr = d->arch.hvm_domain.pci_cf8 + (p->addr & 3); > > + spin_unlock(&d->arch.hvm_domain.pci_lock); > > + } > > + else > > + { > > + type = p->type; > > + addr = p->addr; > > + } > > + > > + rcu_read_lock(&ioreq_server_rcu_lock); > > + > > + switch ( type ) > > + { > > + case IOREQ_TYPE_COPY: > > + case IOREQ_TYPE_PIO: > > + case IOREQ_TYPE_PCI_CONFIG: > > + break; > > + default: > > + goto done; > > + } > > + > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry ) > > + { > > + switch ( type ) > > + { > > + case IOREQ_TYPE_COPY: > > + case IOREQ_TYPE_PIO: { > > + struct list_head *list; > > + struct hvm_io_range *x; > > + > > + list = ( type == IOREQ_TYPE_COPY ) ? > > + &s->mmio_range_list : > > + &s->portio_range_list; > > + > > + list_for_each_entry ( x, > > + list, > > + list_entry ) > > + { > > + if ( (addr >= x->start) && (addr <= x->end) ) > > + goto found; > > + } > > + break; > > + } > > + case IOREQ_TYPE_PCI_CONFIG: { > > + struct hvm_pcidev *x; > > + > > + list_for_each_entry ( x, > > + &s->pcidev_list, > > + list_entry ) > > + { > > + if ( BDF(addr) == x->bdf ) { > > + p->type = type; > > + p->addr = addr; > > + goto found; > > + } > > + } > > + break; > > + } > > + } > > + } > > + > > +done: > > + /* The catch-all server has id 0 */ > > + list_for_each_entry ( s, > > + &d->arch.hvm_domain.ioreq_server_list, > > + list_entry ) > > + { > > + if ( s->id == 0 ) > > + goto found; > > + } > > + > > + s = NULL; > > + > > +found: > > + rcu_read_unlock(&ioreq_server_rcu_lock); > > + > > + return s; > > + > > +#undef BDF > > +} > > + > > int hvm_buffered_io_send(ioreq_t *p) > > { > > struct vcpu *v = current; > > struct domain *d = v->domain; > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > + struct hvm_ioreq_server *s; > > struct hvm_ioreq_page *iorp; > > buffered_iopage_t *pg; > > buf_ioreq_t bp; > > @@ -1660,6 +2228,7 @@ int hvm_buffered_io_send(ioreq_t *p) > > /* Ensure buffered_iopage fits in a page */ > > BUILD_BUG_ON(sizeof(buffered_iopage_t) > PAGE_SIZE); > > > > + s = hvm_select_ioreq_server(v, p); > > if ( !s ) > > return 0; > > > > @@ -1770,18 +2339,34 @@ static bool_t > hvm_send_assist_req_to_server(struct hvm_ioreq_server *s, > > > > bool_t hvm_send_assist_req(struct vcpu *v, ioreq_t *p) > > { > > - struct domain *d = v->domain; > > - struct hvm_ioreq_server *s = d->arch.hvm_domain.ioreq_server; > > + struct hvm_ioreq_server *s; > > > > if ( unlikely(!vcpu_start_shutdown_deferral(v)) ) > > return 0; > > > > + s = hvm_select_ioreq_server(v, p); > > if ( !s ) > > return 0; > > > > return hvm_send_assist_req_to_server(s, v, p); > > } > > > > +void hvm_broadcast_assist_req(struct vcpu *v, ioreq_t *p) > > +{ > > + struct domain *d = v->domain; > > + struct list_head *entry; > > + > > + list_for_each ( entry, > > + &d->arch.hvm_domain.ioreq_server_list ) > > + { > > + struct hvm_ioreq_server *s = list_entry(entry, > > + struct hvm_ioreq_server, > > + list_entry); > > + > > + (void) hvm_send_assist_req_to_server(s, v, p); > > + } > > +} > > + > > void hvm_hlt(unsigned long rflags) > > { > > struct vcpu *curr = current; > > @@ -4370,6 +4955,215 @@ static int hvmop_flush_tlb_all(void) > > return 0; > > } > > > > +static int hvmop_create_ioreq_server( > > + XEN_GUEST_HANDLE_PARAM(xen_hvm_create_ioreq_server_t) uop) > > +{ > > + struct domain *curr_d = current->domain; > > + xen_hvm_create_ioreq_server_t op; > > + struct domain *d; > > + ioservid_t id; > > + int rc; > > + > > + if ( copy_from_guest(&op, uop, 1) ) > > + return -EFAULT; > > + > > + rc = rcu_lock_remote_domain_by_id(op.domid, &d); > > + if ( rc != 0 ) > > + return rc; > > + > > + rc = -EINVAL; > > + if ( !is_hvm_domain(d) ) > > + goto out; > > + > > + rc = -ENOSPC; > > + for ( id = 1; > > + id < d- > >arch.hvm_domain.params[HVM_PARAM_NR_IOREQ_SERVERS]; > > + id++ ) > > + { > > + rc = hvm_create_ioreq_server(d, id, curr_d->domain_id); > > + if ( rc == -EEXIST ) > > + continue; > > + > > + break; > > + } > > + > > + if ( rc == -EEXIST ) > > + rc = -ENOSPC; > > + > > + if ( rc < 0 ) > > + goto out; > > + > > + op.id = id; > > + > > + rc = copy_to_guest(uop, &op, 1) ? -EFAULT : 0; > > + > > +out: > > + rcu_unlock_domain(d); > > + return rc; > > +} > > + > > +static int hvmop_get_ioreq_server_info( > > + XEN_GUEST_HANDLE_PARAM(xen_hvm_get_ioreq_server_info_t) > uop) > > +{ > > + xen_hvm_get_ioreq_server_info_t op; > > + struct domain *d; > > + int rc; > > + > > + if ( copy_from_guest(&op, uop, 1) ) > > + return -EFAULT; > > + > > + rc = rcu_lock_remote_domain_by_id(op.domid, &d); > > + if ( rc != 0 ) > > + return rc; > > + > > + rc = -EINVAL; > > + if ( !is_hvm_domain(d) ) > > + goto out; > > + > > + if ( (rc = hvm_get_ioreq_server_pfn(d, op.id, 0, &op.pfn)) < 0 ) > > + goto out; > > + > > + if ( (rc = hvm_get_ioreq_server_pfn(d, op.id, 1, &op.buf_pfn)) < 0 ) > > + goto out; > > + > > + if ( (rc = hvm_get_ioreq_server_buf_port(d, op.id, &op.buf_port)) < 0 ) > > + goto out; > > + > > + rc = copy_to_guest(uop, &op, 1) ? -EFAULT : 0; > > + > > +out: > > + rcu_unlock_domain(d); > > + return rc; > > +} > > + > > +static int hvmop_map_io_range_to_ioreq_server( > > + > XEN_GUEST_HANDLE_PARAM(xen_hvm_map_io_range_to_ioreq_server_t > ) uop) > > +{ > > + xen_hvm_map_io_range_to_ioreq_server_t op; > > + struct domain *d; > > + int rc; > > + > > + if ( copy_from_guest(&op, uop, 1) ) > > + return -EFAULT; > > + > > + rc = rcu_lock_remote_domain_by_id(op.domid, &d); > > + if ( rc != 0 ) > > + return rc; > > + > > + rc = -EINVAL; > > + if ( !is_hvm_domain(d) ) > > + goto out; > > + > > + rc = hvm_map_io_range_to_ioreq_server(d, op.id, op.is_mmio, > > + op.start, op.end); > > + > > +out: > > + rcu_unlock_domain(d); > > + return rc; > > +} > > + > > +static int hvmop_unmap_io_range_from_ioreq_server( > > + > XEN_GUEST_HANDLE_PARAM(xen_hvm_unmap_io_range_from_ioreq_ser > ver_t) uop) > > +{ > > + xen_hvm_unmap_io_range_from_ioreq_server_t op; > > + struct domain *d; > > + int rc; > > + > > + if ( copy_from_guest(&op, uop, 1) ) > > + return -EFAULT; > > + > > + rc = rcu_lock_remote_domain_by_id(op.domid, &d); > > + if ( rc != 0 ) > > + return rc; > > + > > + rc = -EINVAL; > > + if ( !is_hvm_domain(d) ) > > + goto out; > > + > > + rc = hvm_unmap_io_range_from_ioreq_server(d, op.id, op.is_mmio, > > + op.start); > > + > > +out: > > + rcu_unlock_domain(d); > > + return rc; > > +} > > + > > +static int hvmop_map_pcidev_to_ioreq_server( > > + > XEN_GUEST_HANDLE_PARAM(xen_hvm_map_pcidev_to_ioreq_server_t) > uop) > > +{ > > + xen_hvm_map_pcidev_to_ioreq_server_t op; > > + struct domain *d; > > + int rc; > > + > > + if ( copy_from_guest(&op, uop, 1) ) > > + return -EFAULT; > > + > > + rc = rcu_lock_remote_domain_by_id(op.domid, &d); > > + if ( rc != 0 ) > > + return rc; > > + > > + rc = -EINVAL; > > + if ( !is_hvm_domain(d) ) > > + goto out; > > + > > + rc = hvm_map_pcidev_to_ioreq_server(d, op.id, op.bdf); > > + > > +out: > > + rcu_unlock_domain(d); > > + return rc; > > +} > > + > > +static int hvmop_unmap_pcidev_from_ioreq_server( > > + > XEN_GUEST_HANDLE_PARAM(xen_hvm_unmap_pcidev_from_ioreq_serve > r_t) uop) > > +{ > > + xen_hvm_unmap_pcidev_from_ioreq_server_t op; > > + struct domain *d; > > + int rc; > > + > > + if ( copy_from_guest(&op, uop, 1) ) > > + return -EFAULT; > > + > > + rc = rcu_lock_remote_domain_by_id(op.domid, &d); > > + if ( rc != 0 ) > > + return rc; > > + > > + rc = -EINVAL; > > + if ( !is_hvm_domain(d) ) > > + goto out; > > + > > + rc = hvm_unmap_pcidev_from_ioreq_server(d, op.id, op.bdf); > > + > > +out: > > + rcu_unlock_domain(d); > > + return rc; > > +} > > + > > +static int hvmop_destroy_ioreq_server( > > + XEN_GUEST_HANDLE_PARAM(xen_hvm_destroy_ioreq_server_t) > uop) > > +{ > > + xen_hvm_destroy_ioreq_server_t op; > > + struct domain *d; > > + int rc; > > + > > + if ( copy_from_guest(&op, uop, 1) ) > > + return -EFAULT; > > + > > + rc = rcu_lock_remote_domain_by_id(op.domid, &d); > > + if ( rc != 0 ) > > + return rc; > > + > > + rc = -EINVAL; > > + if ( !is_hvm_domain(d) ) > > + goto out; > > + > > + hvm_destroy_ioreq_server(d, op.id); > > + rc = 0; > > + > > +out: > > + rcu_unlock_domain(d); > > + return rc; > > +} > > + > > long do_hvm_op(unsigned long op, XEN_GUEST_HANDLE_PARAM(void) > arg) > > > > { > > @@ -4378,6 +5172,41 @@ long do_hvm_op(unsigned long op, > XEN_GUEST_HANDLE_PARAM(void) arg) > > > > switch ( op ) > > { > > + case HVMOP_create_ioreq_server: > > + rc = hvmop_create_ioreq_server( > > + guest_handle_cast(arg, xen_hvm_create_ioreq_server_t)); > > + break; > > + > > + case HVMOP_get_ioreq_server_info: > > + rc = hvmop_get_ioreq_server_info( > > + guest_handle_cast(arg, xen_hvm_get_ioreq_server_info_t)); > > + break; > > + > > + case HVMOP_map_io_range_to_ioreq_server: > > + rc = hvmop_map_io_range_to_ioreq_server( > > + guest_handle_cast(arg, > xen_hvm_map_io_range_to_ioreq_server_t)); > > + break; > > + > > + case HVMOP_unmap_io_range_from_ioreq_server: > > + rc = hvmop_unmap_io_range_from_ioreq_server( > > + guest_handle_cast(arg, > xen_hvm_unmap_io_range_from_ioreq_server_t)); > > + break; > > + > > + case HVMOP_map_pcidev_to_ioreq_server: > > + rc = hvmop_map_pcidev_to_ioreq_server( > > + guest_handle_cast(arg, > xen_hvm_map_pcidev_to_ioreq_server_t)); > > + break; > > + > > + case HVMOP_unmap_pcidev_from_ioreq_server: > > + rc = hvmop_unmap_pcidev_from_ioreq_server( > > + guest_handle_cast(arg, > xen_hvm_unmap_pcidev_from_ioreq_server_t)); > > + break; > > + > > + case HVMOP_destroy_ioreq_server: > > + rc = hvmop_destroy_ioreq_server( > > + guest_handle_cast(arg, xen_hvm_destroy_ioreq_server_t)); > > + break; > > + > > case HVMOP_set_param: > > case HVMOP_get_param: > > { > > @@ -4466,9 +5295,9 @@ long do_hvm_op(unsigned long op, > XEN_GUEST_HANDLE_PARAM(void) arg) > > if ( a.value == DOMID_SELF ) > > a.value = curr_d->domain_id; > > > > - rc = hvm_create_ioreq_server(d, a.value); > > + rc = hvm_create_ioreq_server(d, 0, a.value); > > if ( rc == -EEXIST ) > > - rc = hvm_set_ioreq_server_domid(d, a.value); > > + rc = hvm_set_ioreq_server_domid(d, 0, a.value); > > break; > > case HVM_PARAM_ACPI_S_STATE: > > /* Not reflexive, as we must domain_pause(). */ > > @@ -4533,6 +5362,10 @@ long do_hvm_op(unsigned long op, > XEN_GUEST_HANDLE_PARAM(void) arg) > > if ( a.value > SHUTDOWN_MAX ) > > rc = -EINVAL; > > break; > > + case HVM_PARAM_NR_IOREQ_SERVERS: > > + if ( d == current->domain ) > > + rc = -EPERM; > > + break; > > } > > > > if ( rc == 0 ) > > @@ -4567,7 +5400,7 @@ long do_hvm_op(unsigned long op, > XEN_GUEST_HANDLE_PARAM(void) arg) > > case HVM_PARAM_BUFIOREQ_PFN: > > case HVM_PARAM_BUFIOREQ_EVTCHN: > > /* May need to create server */ > > - rc = hvm_create_ioreq_server(d, curr_d->domain_id); > > + rc = hvm_create_ioreq_server(d, 0, curr_d->domain_id); > > if ( rc != 0 && rc != -EEXIST ) > > goto param_fail; > > > > @@ -4576,7 +5409,7 @@ long do_hvm_op(unsigned long op, > XEN_GUEST_HANDLE_PARAM(void) arg) > > case HVM_PARAM_IOREQ_PFN: { > > xen_pfn_t pfn; > > > > - if ( (rc = hvm_get_ioreq_server_pfn(d, 0, &pfn)) < 0 ) > > + if ( (rc = hvm_get_ioreq_server_pfn(d, 0, 0, &pfn)) < > > 0 ) > > goto param_fail; > > > > a.value = pfn; > > @@ -4585,7 +5418,7 @@ long do_hvm_op(unsigned long op, > XEN_GUEST_HANDLE_PARAM(void) arg) > > case HVM_PARAM_BUFIOREQ_PFN: { > > xen_pfn_t pfn; > > > > - if ( (rc = hvm_get_ioreq_server_pfn(d, 1, &pfn)) < 0 ) > > + if ( (rc = hvm_get_ioreq_server_pfn(d, 0, 1, &pfn)) < > > 0 ) > > goto param_fail; > > > > a.value = pfn; > > @@ -4594,7 +5427,7 @@ long do_hvm_op(unsigned long op, > XEN_GUEST_HANDLE_PARAM(void) arg) > > case HVM_PARAM_BUFIOREQ_EVTCHN: { > > evtchn_port_t port; > > > > - if ( (rc = hvm_get_ioreq_server_buf_port(d, &port)) < > > 0 ) > > + if ( (rc = hvm_get_ioreq_server_buf_port(d, 0, &port)) > > < 0 ) > > goto param_fail; > > > > a.value = port; > > diff --git a/xen/arch/x86/hvm/io.c b/xen/arch/x86/hvm/io.c > > index c9adb94..ac0d867 100644 > > --- a/xen/arch/x86/hvm/io.c > > +++ b/xen/arch/x86/hvm/io.c > > @@ -75,7 +75,7 @@ void send_invalidate_req(void) > > .data = ~0UL, /* flush all */ > > }; > > > > - (void)hvm_send_assist_req(v, &p); > > + hvm_broadcast_assist_req(v, &p); > > } > > > > int handle_mmio(void) > > diff --git a/xen/include/asm-x86/hvm/domain.h b/xen/include/asm- > x86/hvm/domain.h > > index a77b83d..e9da543 100644 > > --- a/xen/include/asm-x86/hvm/domain.h > > +++ b/xen/include/asm-x86/hvm/domain.h > > @@ -41,17 +41,38 @@ struct hvm_ioreq_page { > > void *va; > > }; > > > > +struct hvm_io_range { > > + struct list_head list_entry; > > + uint64_t start, end; > > + struct rcu_head rcu; > > +}; > > + > > +struct hvm_pcidev { > > + struct list_head list_entry; > > + uint16_t bdf; > > + struct rcu_head rcu; > > +}; > > + > > struct hvm_ioreq_server { > > + struct list_head list_entry; > > + ioservid_t id; > > struct domain *domain; > > domid_t domid; > > struct hvm_ioreq_page ioreq; > > int ioreq_evtchn[MAX_HVM_VCPUS]; > > struct hvm_ioreq_page buf_ioreq; > > int buf_ioreq_evtchn; > > + struct list_head mmio_range_list; > > + struct list_head portio_range_list; > > + struct list_head pcidev_list; > > }; > > > > struct hvm_domain { > > - struct hvm_ioreq_server *ioreq_server; > > + struct list_head ioreq_server_list; > > + spinlock_t ioreq_server_lock; > > + uint32_t pci_cf8; > > + spinlock_t pci_lock; > > + > > struct pl_time pl_time; > > > > struct hvm_io_handler *io_handler; > > diff --git a/xen/include/asm-x86/hvm/hvm.h b/xen/include/asm- > x86/hvm/hvm.h > > index 40aeddf..4118669 100644 > > --- a/xen/include/asm-x86/hvm/hvm.h > > +++ b/xen/include/asm-x86/hvm/hvm.h > > @@ -229,6 +229,7 @@ int prepare_ring_for_helper(struct domain *d, > unsigned long gmfn, > > void destroy_ring_for_helper(void **_va, struct page_info *page); > > > > bool_t hvm_send_assist_req(struct vcpu *v, ioreq_t *p); > > +void hvm_broadcast_assist_req(struct vcpu *v, ioreq_t *p); > > > > void hvm_get_guest_pat(struct vcpu *v, u64 *guest_pat); > > int hvm_set_guest_pat(struct vcpu *v, u64 guest_pat); > > diff --git a/xen/include/public/hvm/hvm_op.h > b/xen/include/public/hvm/hvm_op.h > > index a9aab4b..6b31189 100644 > > --- a/xen/include/public/hvm/hvm_op.h > > +++ b/xen/include/public/hvm/hvm_op.h > > @@ -23,6 +23,7 @@ > > > > #include "../xen.h" > > #include "../trace.h" > > +#include "../event_channel.h" > > > > /* Get/set subcommands: extra argument == pointer to xen_hvm_param > struct. */ > > #define HVMOP_set_param 0 > > @@ -270,6 +271,75 @@ struct xen_hvm_inject_msi { > > typedef struct xen_hvm_inject_msi xen_hvm_inject_msi_t; > > DEFINE_XEN_GUEST_HANDLE(xen_hvm_inject_msi_t); > > > > +typedef uint32_t ioservid_t; > > + > > +DEFINE_XEN_GUEST_HANDLE(ioservid_t); > > + > > +#define HVMOP_create_ioreq_server 17 > > +struct xen_hvm_create_ioreq_server { > > + domid_t domid; /* IN - domain to be serviced */ > > + ioservid_t id; /* OUT - server id */ > > +}; > > +typedef struct xen_hvm_create_ioreq_server > xen_hvm_create_ioreq_server_t; > > +DEFINE_XEN_GUEST_HANDLE(xen_hvm_create_ioreq_server_t); > > + > > +#define HVMOP_get_ioreq_server_info 18 > > +struct xen_hvm_get_ioreq_server_info { > > + domid_t domid; /* IN - domain to be serviced */ > > + ioservid_t id; /* IN - server id */ > > + xen_pfn_t pfn; /* OUT - ioreq pfn */ > > + xen_pfn_t buf_pfn; /* OUT - buf ioreq pfn */ > > + evtchn_port_t buf_port; /* OUT - buf ioreq port */ > > +}; > > +typedef struct xen_hvm_get_ioreq_server_info > xen_hvm_get_ioreq_server_info_t; > > +DEFINE_XEN_GUEST_HANDLE(xen_hvm_get_ioreq_server_info_t); > > + > > +#define HVMOP_map_io_range_to_ioreq_server 19 > > +struct xen_hvm_map_io_range_to_ioreq_server { > > + domid_t domid; /* IN - domain to be serviced */ > > + ioservid_t id; /* IN - handle from > HVMOP_register_ioreq_server */ > > + int is_mmio; /* IN - MMIO or port IO? */ > > + uint64_aligned_t start, end; /* IN - inclusive start and end of > > range */ > > +}; > > +typedef struct xen_hvm_map_io_range_to_ioreq_server > xen_hvm_map_io_range_to_ioreq_server_t; > > > +DEFINE_XEN_GUEST_HANDLE(xen_hvm_map_io_range_to_ioreq_server_ > t); > > + > > +#define HVMOP_unmap_io_range_from_ioreq_server 20 > > +struct xen_hvm_unmap_io_range_from_ioreq_server { > > + domid_t domid; /* IN - domain to be serviced */ > > + ioservid_t id; /* IN - handle from > > HVMOP_register_ioreq_server */ > > + uint8_t is_mmio; /* IN - MMIO or port IO? */ > > + uint64_aligned_t start; /* IN - start address of the range to remove */ > > +}; > > +typedef struct xen_hvm_unmap_io_range_from_ioreq_server > xen_hvm_unmap_io_range_from_ioreq_server_t; > > > +DEFINE_XEN_GUEST_HANDLE(xen_hvm_unmap_io_range_from_ioreq_se > rver_t); > > + > > +#define HVMOP_map_pcidev_to_ioreq_server 21 > > +struct xen_hvm_map_pcidev_to_ioreq_server { > > + domid_t domid; /* IN - domain to be serviced */ > > + ioservid_t id; /* IN - handle from HVMOP_register_ioreq_server */ > > + uint16_t bdf; /* IN - PCI bus/dev/func */ > > +}; > > +typedef struct xen_hvm_map_pcidev_to_ioreq_server > xen_hvm_map_pcidev_to_ioreq_server_t; > > > +DEFINE_XEN_GUEST_HANDLE(xen_hvm_map_pcidev_to_ioreq_server_t); > > + > > +#define HVMOP_unmap_pcidev_from_ioreq_server 22 > > +struct xen_hvm_unmap_pcidev_from_ioreq_server { > > + domid_t domid; /* IN - domain to be serviced */ > > + ioservid_t id; /* IN - handle from HVMOP_register_ioreq_server */ > > + uint16_t bdf; /* IN - PCI bus/dev/func */ > > +}; > > +typedef struct xen_hvm_unmap_pcidev_from_ioreq_server > xen_hvm_unmap_pcidev_from_ioreq_server_t; > > > +DEFINE_XEN_GUEST_HANDLE(xen_hvm_unmap_pcidev_from_ioreq_serv > er_t); > > + > > +#define HVMOP_destroy_ioreq_server 23 > > +struct xen_hvm_destroy_ioreq_server { > > + domid_t domid; /* IN - domain to be serviced */ > > + ioservid_t id; /* IN - server id */ > > +}; > > +typedef struct xen_hvm_destroy_ioreq_server > xen_hvm_destroy_ioreq_server_t; > > +DEFINE_XEN_GUEST_HANDLE(xen_hvm_destroy_ioreq_server_t); > > + > > #endif /* defined(__XEN__) || defined(__XEN_TOOLS__) */ > > > > #endif /* __XEN_PUBLIC_HVM_HVM_OP_H__ */ > > diff --git a/xen/include/public/hvm/ioreq.h > b/xen/include/public/hvm/ioreq.h > > index f05d130..e84fa75 100644 > > --- a/xen/include/public/hvm/ioreq.h > > +++ b/xen/include/public/hvm/ioreq.h > > @@ -34,6 +34,7 @@ > > > > #define IOREQ_TYPE_PIO 0 /* pio */ > > #define IOREQ_TYPE_COPY 1 /* mmio ops */ > > +#define IOREQ_TYPE_PCI_CONFIG 2 /* pci config ops */ > > #define IOREQ_TYPE_TIMEOFFSET 7 > > #define IOREQ_TYPE_INVALIDATE 8 /* mapcache */ > > > > diff --git a/xen/include/public/hvm/params.h > b/xen/include/public/hvm/params.h > > index 517a184..4109b11 100644 > > --- a/xen/include/public/hvm/params.h > > +++ b/xen/include/public/hvm/params.h > > @@ -145,6 +145,8 @@ > > /* SHUTDOWN_* action in case of a triple fault */ > > #define HVM_PARAM_TRIPLE_FAULT_REASON 31 > > > > -#define HVM_NR_PARAMS 32 > > +#define HVM_PARAM_NR_IOREQ_SERVERS 32 > > + > > +#define HVM_NR_PARAMS 33 > > > > #endif /* __XEN_PUBLIC_HVM_PARAMS_H__ */ _______________________________________________ Xen-devel mailing list Xen-devel@xxxxxxxxxxxxx http://lists.xen.org/xen-devel
|
 |
Lists.xenproject.org is hosted with RackSpace, monitoring our |