On 21.09.2012, at 05:08, David Gibson wrote:
> Below is a patch which implements the (PAPR mandated) NVRAM for the
> pseries machine. It raises a couple of generic questions.
>
> First, this adds a new "nvram" machine option which is used to give a
> block device id to back the NVRAM so it is persistent. Since some
> sort of NVRAM is quite common, it seems this might be useful on other
> machines one day, although obviously nothing else implements it yet.
>
> Second, if a block device is not specified, it simply allocates a
> block of memory to make a non-persistent NVRAM. Obviously that isn't
> really "NV", but it's enough to make many guests happy most of the
> time, and doesn't require setting up an image file and drive. It does
> mean a different set of code paths in the driver though, and it will
> need special case handling for savevm (not implemented yet). Is this
> the right approach, or should I be creating a dummy block device for a
> one-run NVRAM of this kind? I couldn't see an obvious way to do that,
> but maybe I'm missing something.
>
> pseries: Implement PAPR NVRAM
>
> The PAPR specification requires a certain amount of NVRAM, accessed via
> RTAS, which we don't currently implement in qemu. This patch addresses
> this deficiency, implementing the NVRAM as a VIO device, with some glue to
> instantiate it automatically based on a machine option.
>
> The machine option specifies a drive id, which is used to back the NVRAM,
> making it persistent. If nothing is specified, the driver instead simply
> allocates space for the NVRAM, which will not be persistent
>
> Signed-off-by: David Gibson <da...@gibson.dropbear.id.au>
> ---
> hw/ppc/Makefile.objs | 1 +
> hw/spapr.c | 3 +
> hw/spapr.h | 3 +
> hw/spapr_nvram.c | 225 ++++++++++++++++++++++++++++++++++++++++++++++++++
> qemu-config.c | 4 +
> 5 files changed, 236 insertions(+)
> create mode 100644 hw/spapr_nvram.c
>
> diff --git a/hw/ppc/Makefile.objs b/hw/ppc/Makefile.objs
> index 951e407..91cbe8c 100644
> --- a/hw/ppc/Makefile.objs
> +++ b/hw/ppc/Makefile.objs
> @@ -11,6 +11,7 @@ obj-y += ppc_newworld.o
> obj-$(CONFIG_PSERIES) += spapr.o spapr_hcall.o spapr_rtas.o spapr_vio.o
> obj-$(CONFIG_PSERIES) += xics.o spapr_vty.o spapr_llan.o spapr_vscsi.o
> obj-$(CONFIG_PSERIES) += spapr_pci.o pci-hotplug.o spapr_iommu.o
> +obj-$(CONFIG_PSERIES) += spapr_nvram.o
> # PowerPC 4xx boards
> obj-y += ppc4xx_devs.o ppc4xx_pci.o ppc405_uc.o ppc405_boards.o
> obj-y += ppc440_bamboo.o
> diff --git a/hw/spapr.c b/hw/spapr.c
> index a8bd3c1..079825a 100644
> --- a/hw/spapr.c
> +++ b/hw/spapr.c
> @@ -804,6 +804,9 @@ static void ppc_spapr_init(ram_addr_t ram_size,
> }
> }
>
> + /* We always have at least the nvram device on VIO */
> + spapr_create_nvram(spapr);
> +
> /* Set up PCI */
> spapr_pci_rtas_init();
>
> diff --git a/hw/spapr.h b/hw/spapr.h
> index e984e3f..d9c3b4a 100644
> --- a/hw/spapr.h
> +++ b/hw/spapr.h
> @@ -6,11 +6,13 @@
>
> struct VIOsPAPRBus;
> struct sPAPRPHBState;
> +struct sPAPRNVRAM;
> struct icp_state;
>
> typedef struct sPAPREnvironment {
> struct VIOsPAPRBus *vio_bus;
> QLIST_HEAD(, sPAPRPHBState) phbs;
> + struct sPAPRNVRAM *nvram;
> struct icp_state *icp;
>
> target_phys_addr_t ram_limit;
> @@ -336,6 +338,7 @@ typedef struct sPAPRTCE {
> #define SPAPR_PCI_BASE_LIOBN 0x80000000
>
> void spapr_iommu_init(void);
> +void spapr_create_nvram(sPAPREnvironment *spapr);
> DMAContext *spapr_tce_new_dma_context(uint32_t liobn, size_t window_size);
> void spapr_tce_free(DMAContext *dma);
> void spapr_tce_reset(DMAContext *dma);
> diff --git a/hw/spapr_nvram.c b/hw/spapr_nvram.c
> new file mode 100644
> index 0000000..8cd8a53
> --- /dev/null
> +++ b/hw/spapr_nvram.c
> @@ -0,0 +1,225 @@
> +/*
> + * QEMU sPAPR NVRAM emulation
> + *
> + * Copyright (C) 2012 David Gibson, IBM Corporation.
> + *
> + * Permission is hereby granted, free of charge, to any person obtaining a
> copy
> + * of this software and associated documentation files (the "Software"), to
> deal
> + * in the Software without restriction, including without limitation the
> rights
> + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
> + * copies of the Software, and to permit persons to whom the Software is
> + * furnished to do so, subject to the following conditions:
> + *
> + * The above copyright notice and this permission notice shall be included in
> + * all copies or substantial portions of the Software.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
> FROM,
> + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
> + * THE SOFTWARE.
> + */
> +#include <sys/mman.h>
> +#include <libfdt.h>
> +
> +#include "device_tree.h"
> +#include "hw/sysbus.h"
> +#include "hw/spapr.h"
> +#include "hw/spapr_vio.h"
> +
> +typedef struct sPAPRNVRAM {
> + VIOsPAPRDevice sdev;
> + uint32_t size;
> + uint8_t *buf;
> + BlockDriverState *drive;
> +} sPAPRNVRAM;
> +
> +#define MIN_NVRAM_SIZE 8192
> +#define DEFAULT_NVRAM_SIZE 16384
> +#define MAX_NVRAM_SIZE (UINT16_MAX * 16)
> +
> +static void rtas_nvram_fetch(sPAPREnvironment *spapr,
> + uint32_t token, uint32_t nargs,
> + target_ulong args,
> + uint32_t nret, target_ulong rets)
> +{
> + sPAPRNVRAM *nvram = spapr->nvram;
> + target_phys_addr_t offset, buffer, len;
> + int alen;
> + void *membuf;
> +
> + if ((nargs != 3) || (nret != 2)) {
> + rtas_st(rets, 0, -3);
> + return;
> + }
> +
> + if (!nvram) {
> + rtas_st(rets, 0, -1);
> + rtas_st(rets, 1, 0);
> + return;
> + }
> +
> + offset = rtas_ld(args, 0);
> + buffer = rtas_ld(args, 1);
> + len = rtas_ld(args, 2);
> +
> + if (((offset + len) < offset)
> + || ((offset + len) > nvram->size)) {
> + rtas_st(rets, 0, -3);
> + rtas_st(rets, 1, 0);
> + return;
> + }
> +
> + membuf = cpu_physical_memory_map(buffer, &len, 1);
> + if (nvram->drive) {
> + alen = bdrv_pread(nvram->drive, offset, membuf, len);
> + } else {
> + assert(nvram->buf);
> +
> + memcpy(membuf, nvram->buf + offset, len);
> + alen = len;
> + }
> + cpu_physical_memory_unmap(membuf, len, 1, len);
> +
> + rtas_st(rets, 0, (alen < len) ? -1 : 0);
> + rtas_st(rets, 1, (alen < 0) ? 0 : alen);
> +}
> +
> +static void rtas_nvram_store(sPAPREnvironment *spapr,
> + uint32_t token, uint32_t nargs,
> + target_ulong args,
> + uint32_t nret, target_ulong rets)
> +{
> + sPAPRNVRAM *nvram = spapr->nvram;
> + target_phys_addr_t offset, buffer, len;
> + int alen;
> + void *membuf;
> +
> + if ((nargs != 3) || (nret != 2)) {
> + rtas_st(rets, 0, -3);
> + return;
> + }
> +
> + if (!nvram) {
> + rtas_st(rets, 0, -1);
> + return;
> + }
> +
> + offset = rtas_ld(args, 0);
> + buffer = rtas_ld(args, 1);
> + len = rtas_ld(args, 2);
> +
> + if (((offset + len) < offset)
> + || ((offset + len) > nvram->size)) {
> + rtas_st(rets, 0, -3);
> + return;
> + }
> +
> + membuf = cpu_physical_memory_map(buffer, &len, 0);
> + if (nvram->drive) {
> + alen = bdrv_pwrite(nvram->drive, offset, membuf, len);
> + } else {
> + assert(nvram->buf);
> +
> + memcpy(nvram->buf + offset, membuf, len);
> + alen = len;
> + }
> + cpu_physical_memory_unmap(membuf, len, 0, len);
> +
> + rtas_st(rets, 0, (alen < len) ? -1 : 0);
> + rtas_st(rets, 1, (alen < 0) ? 0 : alen);
> +}
> +
> +static int spapr_nvram_init(VIOsPAPRDevice *dev)
> +{
> + sPAPRNVRAM *nvram = (sPAPRNVRAM *)dev;
> +
> + if (nvram->drive) {
> + nvram->size = bdrv_getlength(nvram->drive);
> + } else {
> + nvram->size = DEFAULT_NVRAM_SIZE;
> + nvram->buf = g_malloc0(nvram->size);
> + }
> +
> + if ((nvram->size < MIN_NVRAM_SIZE) || (nvram->size > MAX_NVRAM_SIZE)) {
> + fprintf(stderr, "spapr-nvram must be between %d and %d bytes in
> size\n",
> + MIN_NVRAM_SIZE, MAX_NVRAM_SIZE);
> + return -1;
> + }
> +
> + return 0;
> +}
> +
> +static int spapr_nvram_devnode(VIOsPAPRDevice *dev, void *fdt, int node_off)
> +{
> + sPAPRNVRAM *nvram = (sPAPRNVRAM *)dev;
> +
> + return fdt_setprop_cell(fdt, node_off, "#bytes", nvram->size);
> +}
> +
> +static Property spapr_nvram_properties[] = {
> + DEFINE_SPAPR_PROPERTIES(sPAPRNVRAM, sdev),
> + DEFINE_PROP_DRIVE("drive", sPAPRNVRAM, drive),
> + DEFINE_PROP_END_OF_LIST(),
> +};
> +
> +static void spapr_nvram_class_init(ObjectClass *klass, void *data)
> +{
> + DeviceClass *dc = DEVICE_CLASS(klass);
> + VIOsPAPRDeviceClass *k = VIO_SPAPR_DEVICE_CLASS(klass);
> +
> + k->init = spapr_nvram_init;
> + k->devnode = spapr_nvram_devnode;
> + k->dt_name = "nvram";
> + k->dt_type = "nvram";
> + k->dt_compatible = "qemu,spapr-nvram";
> + dc->props = spapr_nvram_properties;
> +}
> +
> +static const TypeInfo spapr_nvram_type_info = {
> + .name = "spapr-nvram",
> + .parent = TYPE_VIO_SPAPR_DEVICE,
> + .instance_size = sizeof(sPAPRNVRAM),
> + .class_init = spapr_nvram_class_init,
> +};
> +
> +static void spapr_nvram_register_types(void)
> +{
> + type_register_static(&spapr_nvram_type_info);
> +}
> +
> +type_init(spapr_nvram_register_types)
> +
> +void spapr_create_nvram(sPAPREnvironment *spapr)
> +{
> + QemuOpts *machine_opts;
> + DeviceState *dev;
> +
> + dev = qdev_create(&spapr->vio_bus->bus, "spapr-nvram");
> +
> + machine_opts = qemu_opts_find(qemu_find_opts("machine"), 0);
> + if (machine_opts) {
> + const char *drivename;
> +
> + drivename = qemu_opt_get(machine_opts, "nvram");
> + if (drivename) {
> + BlockDriverState *bs;
> +
> + bs = bdrv_find(drivename);
> + if (!bs) {
> + fprintf(stderr, "No such block device \"%s\" for nvram\n",
> + drivename);
> + exit(1);
I thought you wanted to autogenerate contents in RAM if no drive is available?
> + }
> + qdev_prop_set_drive_nofail(dev, "drive", bs);
> + }
> + }
> +
> + qdev_init_nofail(dev);
> +
> + spapr->nvram = (sPAPRNVRAM *)dev;
> + spapr_rtas_register("nvram-fetch", rtas_nvram_fetch);
> + spapr_rtas_register("nvram-store", rtas_nvram_store);
> +}
> diff --git a/qemu-config.c b/qemu-config.c
> index 12eafbb..1cd9a1b 100644
> --- a/qemu-config.c
> +++ b/qemu-config.c
> @@ -619,6 +619,10 @@ static QemuOptsList qemu_machine_opts = {
> .name = "mem-merge",
> .type = QEMU_OPT_BOOL,
> .help = "enable/disable memory merge support",
> + }, {
> + .name = "nvram",
> + .type = QEMU_OPT_STRING,
> + .help = "Drive backing persistent NVRAM",
I like the idea of a machine implemented NVRAM. Maybe we should add an -nvram
option that does an automatic -drive and -machine ...,machine=drive_id for the
user :).
Alex
