July 2012 - SeaBIOS - mail.coreboot.org

Re: [SeaBIOS] boot.c: handle_18, handle_19, do_boot
by Kevin O'Connor July 12, 2012

July 12, 2012

On Wed, Jul 11, 2012 at 09:17:04AM -0600, Steve Goodrich wrote: > Just a quick question: do these functions (handle_18, handle_19, and do_boot > from boot.c) run in 16-bit mode or in 32-bit mode? The boot phase runs in 32bit mode. See entry_18/entry_19 in romlayout.S. -Kevin

1 0

Re: [SeaBIOS] [RFC PATCH v2 20/21] Implement -dimms, -dimmspop command line options
by Avi Kivity July 11, 2012

July 11, 2012

On 07/11/2012 01:32 PM, Vasilis Liaskovitis wrote: > Implement batch dimm creation command line options. These could be useful for > not bloating the command line with a large number of dimms. IMO this is unneeded. With a management tool there is no problem generating a long command line; from the command line -dimm will be a rarely used option. > > syntax: -dimms pfx=poolid,size=sz,num=n > Will create numdimms dimms with ids poolid0, ..., poolidn-1. Each dimm has a > size of sz. > > Implement -dimmpop option to populate dimms at bootup > syntax: -dimmpop pfx=poolid,num=n > This will populate n dimms with ids poolid0, ..., poolidn-1. > > (live-migration could break here without patch 12/21: -dimmspop > needs to be reworked to support populating of individual dimms with > same prefix, and not only a range of dimms starting from 0) > -- error compiling committee.c: too many arguments to function

2 1

Re: [SeaBIOS] [Qemu-devel] [RFC PATCH v2 19/21] Implement "info memtotal" and "query-memtotal"
by Eric Blake July 11, 2012

July 11, 2012

On 07/11/2012 04:32 AM, Vasilis Liaskovitis wrote: > Returns total memory of guest in bytes, including hotplugged memory. > > Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> Should this instead be merged with query-balloon output, so that we have a single command that shows all aspects of memory usage (both balloon and hotplug at once)? > @@ -1888,3 +1888,15 @@ > # Since: 1.1.3 > ## > { 'command': 'query-memhp', 'returns': ['MemHpInfo'] } > + > +## > +# @query-memtotal: A more generic name might be 'query-memory', especially if we merge balloon and hotplug information into one command. > +# > +# Returns total memory in bytes, including hotplugged dimms > +# > +# Returns: a l truncated -- Eric Blake eblake(a)redhat.com +1-919-301-3266 Libvirt virtualization library http://libvirt.org

2 1

Re: [SeaBIOS] [Qemu-devel] [RFC PATCH v2 13/21] Implement memory hotplug notification lists
by Eric Blake July 11, 2012

July 11, 2012

On 07/11/2012 04:31 AM, Vasilis Liaskovitis wrote: > Guest can respond to ACPI hotplug events e.g. with _EJ or _OST method. > This patch implements a tail queue to store guest notifications for memory > hot-add and hot-remove requests. > > Guest responses for memory hotplug command on a per-dimm basis can be detected > with the new hmp command "info memhp" or the new qmp command "query-memhp" > Examples: > > +++ b/qapi-schema.json > @@ -1862,3 +1862,29 @@ > # Since: 0.14.0 > ## > { 'command': 'netdev_del', 'data': {'id': 'str'} } > + > +## > +# @MemHpInfo: > +# > +# Information about status of a memory hotplug command > +# > +# @Dimm: the Dimm associated with the result > +# > +# @result: the result of the hotplug command > +# > +# Since: 1.1.3 Should probably be 1.2, not 1.1.3. > +# > +## > +{ 'type': 'MemHpInfo', > + 'data': {'Dimm': 'str', 'request': 'str', 'result': 'str'} } Why the upper case? Wouldn't 'dimm' be more consistent? > + > +## > +# @query-memhp: Why are we abbreviating? It might be better to name the QMP command query-memory-hotplug > +# > +# Returns a list of information about pending hotplug commands > +# > +# Returns: a list of @MemhpInfo > +# > +# Since: 1.1.3 Likewise for 1.2. > + > +- "Dimm": Dimm name (json-str) > +- "request": type of hot request: hot-add or hot-remove (json-str) > +- "result": result of the hotplug request for this Dimm success or failure (json-str) This may need tweaks (such as s/Dimm/dimm/) based on resolution of above comments. -- Eric Blake eblake(a)redhat.com +1-919-301-3266 Libvirt virtualization library http://libvirt.org

2 1

Re: [SeaBIOS] [RFC PATCH v2 05/21] pciinit: Fix pcimem_start value
by Gerd Hoffmann July 11, 2012

July 11, 2012

On 07/11/12 12:31, Vasilis Liaskovitis wrote: > In order to hotplug memory between RamSize and BUILD_PCIMEM_START, the pci > window needs to start at BUILD_PCIMEM_START (0xe0000000). > Otherwise, the guest cannot online new dimms at those ranges due to pci_root > window conflicts. (workaround for linux guest is booting with pci=nocrs) > static void pci_bios_map_devices(struct pci_bus *busses) > { > - pcimem_start = RamSize; > + pcimem_start = BUILD_PCIMEM_START; It isn't that simple. For the 32bit pci window it will work, but will leaves address space unused instead of assigning it to the 32bit pci window. For the 64bit pci window it will not work. You have to walk the dimms and figure what the highest used address is, for both below-4g and above-4g. Then fill two variable with it and make the pci init code use that instead of RamSize and RamSizeOver4G. cheers, Gerd

2 1

Re: [SeaBIOS] [RFC PATCH v2 04/21] acpi: generate hotplug memory devices
by Wen Congyang July 11, 2012

July 11, 2012

At 07/11/2012 06:31 PM, Vasilis Liaskovitis Wrote: > The memory device generation is guided by qemu paravirt info. Seabios > first uses the info to setup SRAT entries for the hotplug-able memory slots. > Afterwards, build_memssdt uses the created SRAT entries to generate > appropriate memory device objects. One memory device (and corresponding SRAT > entry) is generated for each hotplug-able qemu memslot. Currently no SSDT > memory device is created for initial system memory. > > We only support up to 255 DIMMs for now (PackageOp used for the MEON array can > only describe an array of at most 255 elements. VarPackageOp would be needed to > support more than 255 devices) > > v1->v2: > Seabios reads mems_sts from qemu to build e820_map > SSDT size and some offsets are calculated with extraction macros. > > Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> > --- > src/acpi.c | 158 +++++++++++++++++++++++++++++++++++++++++++++++++++++++++-- > 1 files changed, 152 insertions(+), 6 deletions(-) > > diff --git a/src/acpi.c b/src/acpi.c > index 55e4607..c83e8c7 100644 > --- a/src/acpi.c > +++ b/src/acpi.c > @@ -510,6 +510,127 @@ build_ssdt(void) > return ssdt; > } > > +#include "ssdt-mem.hex" > + > +/* 0x5B 0x82 DeviceOp PkgLength NameString DimmID */ > +#define MEM_BASE 0xaf80 > +#define SD_MEM (ssdm_mem_aml + *ssdt_mem_start) > +#define SD_MEMSIZEOF (*ssdt_mem_end - *ssdt_mem_start) > +#define SD_OFFSET_MEMHEX (*ssdt_mem_name - *ssdt_mem_start + 2) > +#define SD_OFFSET_MEMID (*ssdt_mem_id - *ssdt_mem_start) > +#define SD_OFFSET_PXMID 31 > +#define SD_OFFSET_MEMSTART 55 > +#define SD_OFFSET_MEMEND 63 > +#define SD_OFFSET_MEMSIZE 79 > + > +u64 nb_hp_memslots = 0; > +struct srat_memory_affinity *mem; > + > +static void build_memdev(u8 *ssdt_ptr, int i, u64 mem_base, u64 mem_len, u8 node) > +{ > + memcpy(ssdt_ptr, SD_MEM, SD_MEMSIZEOF); > + ssdt_ptr[SD_OFFSET_MEMHEX] = getHex(i >> 4); > + ssdt_ptr[SD_OFFSET_MEMHEX+1] = getHex(i); > + ssdt_ptr[SD_OFFSET_MEMID] = i; > + ssdt_ptr[SD_OFFSET_PXMID] = node; > + *(u64*)(ssdt_ptr + SD_OFFSET_MEMSTART) = mem_base; > + *(u64*)(ssdt_ptr + SD_OFFSET_MEMEND) = mem_base + mem_len; > + *(u64*)(ssdt_ptr + SD_OFFSET_MEMSIZE) = mem_len; > +} > + > +static void* > +build_memssdt(void) > +{ > + u64 mem_base; > + u64 mem_len; > + u8 node; > + int i; > + struct srat_memory_affinity *entry = mem; > + u64 nb_memdevs = nb_hp_memslots; > + u8 memslot_status, enabled; > + > + int length = ((1+3+4) > + + (nb_memdevs * SD_MEMSIZEOF) > + + (1+2+5+(12*nb_memdevs)) > + + (6+2+1+(1*nb_memdevs))); > + u8 *ssdt = malloc_high(sizeof(struct acpi_table_header) + length); > + if (! ssdt) { > + warn_noalloc(); > + return NULL; > + } > + u8 *ssdt_ptr = ssdt + sizeof(struct acpi_table_header); > + > + // build Scope(_SB_) header > + *(ssdt_ptr++) = 0x10; // ScopeOp > + ssdt_ptr = encodeLen(ssdt_ptr, length-1, 3); > + *(ssdt_ptr++) = '_'; > + *(ssdt_ptr++) = 'S'; > + *(ssdt_ptr++) = 'B'; > + *(ssdt_ptr++) = '_'; > + > + for (i = 0; i < nb_memdevs; i++) { > + mem_base = (((u64)(entry->base_addr_high) << 32 )| entry->base_addr_low); > + mem_len = (((u64)(entry->length_high) << 32 )| entry->length_low); > + node = entry->proximity[0]; > + build_memdev(ssdt_ptr, i, mem_base, mem_len, node); > + ssdt_ptr += SD_MEMSIZEOF; > + entry++; > + } > + > + // build "Method(MTFY, 2) {If (LEqual(Arg0, 0x00)) {Notify(CM00, Arg1)} ...}" > + *(ssdt_ptr++) = 0x14; // MethodOp > + ssdt_ptr = encodeLen(ssdt_ptr, 2+5+(12*nb_memdevs), 2); > + *(ssdt_ptr++) = 'M'; > + *(ssdt_ptr++) = 'T'; > + *(ssdt_ptr++) = 'F'; > + *(ssdt_ptr++) = 'Y'; > + *(ssdt_ptr++) = 0x02; > + for (i=0; i<nb_memdevs; i++) { > + *(ssdt_ptr++) = 0xA0; // IfOp > + ssdt_ptr = encodeLen(ssdt_ptr, 11, 1); > + *(ssdt_ptr++) = 0x93; // LEqualOp > + *(ssdt_ptr++) = 0x68; // Arg0Op > + *(ssdt_ptr++) = 0x0A; // BytePrefix > + *(ssdt_ptr++) = i; > + *(ssdt_ptr++) = 0x86; // NotifyOp > + *(ssdt_ptr++) = 'M'; > + *(ssdt_ptr++) = 'P'; > + *(ssdt_ptr++) = getHex(i >> 4); > + *(ssdt_ptr++) = getHex(i); > + *(ssdt_ptr++) = 0x69; // Arg1Op > + } > + > + // build "Name(MEON, Package() { One, One, ..., Zero, Zero, ... })" > + *(ssdt_ptr++) = 0x08; // NameOp > + *(ssdt_ptr++) = 'M'; > + *(ssdt_ptr++) = 'E'; > + *(ssdt_ptr++) = 'O'; > + *(ssdt_ptr++) = 'N'; > + *(ssdt_ptr++) = 0x12; // PackageOp > + ssdt_ptr = encodeLen(ssdt_ptr, 2+1+(1*nb_memdevs), 2); > + *(ssdt_ptr++) = nb_memdevs; > + > + entry = mem; > + memslot_status = 0; > + > + for (i = 0; i < nb_memdevs; i++) { > + enabled = 0; > + if (i % 8 == 0) > + memslot_status = inb(MEM_BASE + i/8); > + enabled = memslot_status & 1; > + mem_base = (((u64)(entry->base_addr_high) << 32 )| entry->base_addr_low); > + mem_len = (((u64)(entry->length_high) << 32 )| entry->length_low); > + *(ssdt_ptr++) = enabled ? 0x01 : 0x00; > + if (enabled) > + add_e820(mem_base, mem_len, E820_RAM); add_e820() is declared in memmap.h. You should include this header file, otherwise, seabios cannot be built. Thanks Wen Congyang > + memslot_status = memslot_status >> 1; > + entry++; > + } > + build_header((void*)ssdt, SSDT_SIGNATURE, ssdt_ptr - ssdt, 1); > + > + return ssdt; > +} > + > #include "ssdt-pcihp.hex" > > #define PCI_RMV_BASE 0xae0c > @@ -618,9 +739,6 @@ build_srat(void) > { > int nb_numa_nodes = qemu_cfg_get_numa_nodes(); > > - if (nb_numa_nodes == 0) > - return NULL; > - > u64 *numadata = malloc_tmphigh(sizeof(u64) * (MaxCountCPUs + nb_numa_nodes)); > if (!numadata) { > warn_noalloc(); > @@ -629,10 +747,11 @@ build_srat(void) > > qemu_cfg_get_numa_data(numadata, MaxCountCPUs + nb_numa_nodes); > > + qemu_cfg_get_numa_data(&nb_hp_memslots, 1); > struct system_resource_affinity_table *srat; > int srat_size = sizeof(*srat) + > sizeof(struct srat_processor_affinity) * MaxCountCPUs + > - sizeof(struct srat_memory_affinity) * (nb_numa_nodes + 2); > + sizeof(struct srat_memory_affinity) * (nb_numa_nodes + nb_hp_memslots + 2); > > srat = malloc_high(srat_size); > if (!srat) { > @@ -667,7 +786,7 @@ build_srat(void) > * from 640k-1M and possibly another one from 3.5G-4G. > */ > struct srat_memory_affinity *numamem = (void*)core; > - int slots = 0; > + int slots = 0, node; > u64 mem_len, mem_base, next_base = 0; > > acpi_build_srat_memory(numamem, 0, 640*1024, 0, 1); > @@ -694,10 +813,36 @@ build_srat(void) > next_base += (1ULL << 32) - RamSize; > } > acpi_build_srat_memory(numamem, mem_base, mem_len, i-1, 1); > + > numamem++; > slots++; > + > + } > + mem = (void*)numamem; > + > + if (nb_hp_memslots) { > + u64 *hpmemdata = malloc_tmphigh(sizeof(u64) * (3 * nb_hp_memslots)); > + if (!hpmemdata) { > + warn_noalloc(); > + free(hpmemdata); > + free(numadata); > + return NULL; > + } > + > + qemu_cfg_get_numa_data(hpmemdata, 3 * nb_hp_memslots); > + > + for (i = 1; i < nb_hp_memslots + 1; ++i) { > + mem_base = *hpmemdata++; > + mem_len = *hpmemdata++; > + node = *hpmemdata++; > + acpi_build_srat_memory(numamem, mem_base, mem_len, node, 1); > + numamem++; > + slots++; > + } > + free(hpmemdata); > } > - for (; slots < nb_numa_nodes + 2; slots++) { > + > + for (; slots < nb_numa_nodes + nb_hp_memslots + 2; slots++) { > acpi_build_srat_memory(numamem, 0, 0, 0, 0); > numamem++; > } > @@ -748,6 +893,7 @@ acpi_bios_init(void) > ACPI_INIT_TABLE(build_madt()); > ACPI_INIT_TABLE(build_hpet()); > ACPI_INIT_TABLE(build_srat()); > + ACPI_INIT_TABLE(build_memssdt()); > ACPI_INIT_TABLE(build_pcihp()); > > u16 i, external_tables = qemu_cfg_acpi_additional_tables();

2 1

boot.c: handle_18, handle_19, do_boot
by Steve Goodrich July 11, 2012

July 11, 2012

Just a quick question: do these functions (handle_18, handle_19, and do_boot from boot.c) run in 16-bit mode or in 32-bit mode? Thanks! -- Steve G.

1 0

[RFC PATCH v2 21/21] Implement mem_increase, mem_decrease hmp/qmp commands
by Vasilis Liaskovitis July 11, 2012

July 11, 2012

This implements batch monitor operations for hot-add and hot-remove. These are probably better suited for a higher-level management layer, but are useful for testing. Let me know if there is interest for such commands upstream. syntax: mem_increase poolid num will hotplug num dimms from pool poolid. This starts from lowest unpopulated physical memory (dimm) and trying to cover any existing physical holes. syntax: mem_decrease poolid num will hot-unplug num dimms from pool poolid, This starts from highest populated physical memory (dimm). Respective qmp commands are "mem-increase", "mem-decrease". Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> --- hmp-commands.hx | 31 ++++++++++++++++ hw/dimm.c | 104 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ hw/dimm.h | 7 ++++ monitor.c | 10 +++++ monitor.h | 2 + qmp-commands.hx | 40 +++++++++++++++++++++ 6 files changed, 194 insertions(+), 0 deletions(-) diff --git a/hmp-commands.hx b/hmp-commands.hx index 016062e..e0c1cf4 100644 --- a/hmp-commands.hx +++ b/hmp-commands.hx @@ -648,6 +648,37 @@ STEXI Hot-add dimm. ETEXI + { + .name = "mem_increase", + .args_type = "pfx:s,num:s", + .params = "pfx num", + .help = "hot-plug num dimms of memory pool pfx", + .user_print = monitor_user_noop, + .mhandler.cmd_new = do_dimm_add_range, + }, + +STEXI +@item mem_increase @var{config} +@findex mem_increase + +Hotplug dimms. +ETEXI + + { + .name = "mem_decrease", + .args_type = "pfx:s,num:s", + .params = "pfx num", + .help = "hot-unplug num dimms of memory pool pfx", + .user_print = monitor_user_noop, + .mhandler.cmd_new = do_dimm_del_range, + }, + +STEXI +@item mem_decrease @var{config} +@findex mem_decrease + +Hot-unplug dimms. +ETEXI { .name = "device_del", diff --git a/hw/dimm.c b/hw/dimm.c index b544173..48542ba 100644 --- a/hw/dimm.c +++ b/hw/dimm.c @@ -166,6 +166,110 @@ int dimm_do(Monitor *mon, const QDict *qdict, bool add) return 0; } +/* Find for dimm_do_range operation + DIMM_MIN_UNPOPULATED: used for finding next DIMM to hotplug + DIMM_MAX_POPULATED: used for finding next DIMM for hot-unplug + */ + +DimmState *dimm_find_next(char *pfx, uint32_t mode) +{ + DeviceState *dev; + DimmState *slot, *ret; + const char *type; + uint32_t idx; + + Error *err = NULL; + BusChild *kid; + BusState *bus = sysbus_get_default(); + ret = NULL; + + if (mode == DIMM_MIN_UNPOPULATED) + idx = MAX_DIMMS; + else if (mode == DIMM_MAX_POPULATED) + idx = 0; + else + return false; + + QTAILQ_FOREACH(kid, &bus->children, sibling) { + dev = kid->child; + type = object_property_get_str(OBJECT(dev), "type", &err); + if (err) { + error_free(err); + fprintf(stderr, "error getting device type\n"); + exit(1); + } + + if (!strcmp(type, "dimm")) { + slot = DIMM(dev); + if (strstr(dev->id, pfx) && strcmp(dev->id, pfx)) { + if (mode == DIMM_MIN_UNPOPULATED && + (slot->populated == false) && + (slot->pending == false) && + (idx > slot->idx)) { + idx = slot->idx; + ret = slot; + } + else if (mode == DIMM_MAX_POPULATED && + (slot->populated == true) && + (slot->pending == false) && + (idx <= slot->idx)) { + idx = slot->idx; + ret = slot; + } + } + } + } + return ret; +} + +int dimm_do_range(Monitor *mon, const QDict *qdict, bool add) +{ + DimmState *slot = NULL; + uint32_t mode; + uint32_t idx; + int num, ndimms; + + char *pfx = (char*) qdict_get_try_str(qdict, "pfx"); + if (!pfx) { + fprintf(stderr, "ERROR %s invalid pfx\n",__FUNCTION__); + return 1; + } + + char *value = (char*) qdict_get_try_str(qdict, "num"); + if (!value) { + fprintf(stderr, "ERROR %s invalid pfx\n",__FUNCTION__); + return 1; + } + num = atoi(value); + + if (add) + mode = DIMM_MIN_UNPOPULATED; + else + mode = DIMM_MAX_POPULATED; + + ndimms = 0; + while (ndimms < num) { + slot = dimm_find_next(pfx, mode); + if (slot == NULL) { + fprintf(stderr, "%s no further slot found for pool %s\n", + __FUNCTION__, pfx); + fprintf(stderr, "%s operated on %d / %d requested dimms\n", + __FUNCTION__, ndimms, num); + return 1; + } + + if (add) { + dimm_activate(slot); + } + else { + dimm_deactivate(slot); + } + ndimms++; + idx++; + } + + return 0; +} DimmState *dimm_find_from_idx(uint32_t idx) { DimmState *slot; diff --git a/hw/dimm.h b/hw/dimm.h index 0fdf59b..7c456fa 100644 --- a/hw/dimm.h +++ b/hw/dimm.h @@ -11,6 +11,11 @@ #define DIMM_BITMAP_BYTES (MAX_DIMMS + 7) / 8 #define DEFAULT_DIMMSIZE 1024*1024*1024 +enum { + DIMM_MIN_UNPOPULATED= 0, + DIMM_MAX_POPULATED = 1 +}; + typedef enum { DIMM_REMOVE_SUCCESS = 0, DIMM_REMOVE_FAIL = 1, @@ -61,5 +66,7 @@ void dimm_deactivate(DimmState *slot); void dimm_scan_populated(void); void dimm_notify(uint32_t idx, uint32_t event); int dimm_set_populated(DimmState *s); +DimmState *dimm_find_next(char *pfx, uint32_t mode); +int dimm_do_range(Monitor *mon, const QDict *qdict, bool add); #endif diff --git a/monitor.c b/monitor.c index 1dd646c..2e0ce1f 100644 --- a/monitor.c +++ b/monitor.c @@ -4838,3 +4838,13 @@ int do_dimm_del(Monitor *mon, const QDict *qdict, QObject **ret_data) { return dimm_do(mon, qdict, false); } + +int do_dimm_add_range(Monitor *mon, const QDict *qdict, QObject **ret_data) +{ + return dimm_do_range(mon, qdict, true); +} + +int do_dimm_del_range(Monitor *mon, const QDict *qdict, QObject **ret_data) +{ + return dimm_do_range(mon, qdict, false); +} diff --git a/monitor.h b/monitor.h index afdd721..8224301 100644 --- a/monitor.h +++ b/monitor.h @@ -88,5 +88,7 @@ int qmp_qom_get(Monitor *mon, const QDict *qdict, QObject **ret); int do_dimm_add(Monitor *mon, const QDict *qdict, QObject **ret_data); int do_dimm_del(Monitor *mon, const QDict *qdict, QObject **ret_data); +int do_dimm_add_range(Monitor *mon, const QDict *qdict, QObject **ret_data); +int do_dimm_del_range(Monitor *mon, const QDict *qdict, QObject **ret_data); #endif /* !MONITOR_H */ diff --git a/qmp-commands.hx b/qmp-commands.hx index 6c71696..c3f74ea 100644 --- a/qmp-commands.hx +++ b/qmp-commands.hx @@ -2306,3 +2306,43 @@ Example: } EQMP + + { + .name = "mem-increase", + .args_type = "pfx:s,num:s", + .mhandler.cmd_new = do_dimm_add_range, + }, +SQMP +mem-increase +------------- + +Hotplug memory DIMMs from memory pool + +Will hotplug num memory DIMMs from pool with name pfx. + +Example: + +-> { "execute": "mem-increase", "arguments": { "pfx" : "pool", "num": "10" } } +<- { "return": {} } + +EQMP + + { + .name = "mem-decrease", + .args_type = "pfx:s,num:s", + .mhandler.cmd_new = do_dimm_del_range, + }, +SQMP +mem-decrease +------------- + +Hot-unplug memory DIMMs from memory pool + +Will hot-unplug num memory DIMMs from pool with name pfx. + +Example: + +-> { "execute": "mem-decrease", "arguments": { "pfx" : "pool", "num": "10" } } +<- { "return": {} } + +EQMP -- 1.7.9

1 0

[RFC PATCH v2 20/21] Implement -dimms, -dimmspop command line options
by Vasilis Liaskovitis July 11, 2012

July 11, 2012

Implement batch dimm creation command line options. These could be useful for not bloating the command line with a large number of dimms. syntax: -dimms pfx=poolid,size=sz,num=n Will create numdimms dimms with ids poolid0, ..., poolidn-1. Each dimm has a size of sz. Implement -dimmpop option to populate dimms at bootup syntax: -dimmpop pfx=poolid,num=n This will populate n dimms with ids poolid0, ..., poolidn-1. (live-migration could break here without patch 12/21: -dimmspop needs to be reworked to support populating of individual dimms with same prefix, and not only a range of dimms starting from 0) Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> --- hw/dimm.c | 9 ++++++ hw/dimm.h | 2 +- qemu-config.c | 45 ++++++++++++++++++++++++++++ qemu-options.hx | 10 ++++++ vl.c | 86 ++++++++++++++++++++++++++++++++++++++++++++++++++++++- 5 files changed, 150 insertions(+), 2 deletions(-) diff --git a/hw/dimm.c b/hw/dimm.c index 2115567..6e324d3 100644 --- a/hw/dimm.c +++ b/hw/dimm.c @@ -187,6 +187,15 @@ void dimm_calc_offsets(dimm_calcoffset_fn calcfn) } } +int dimm_set_populated(DimmState *s) +{ + if (s) { + s->populated = true; + return 0; + } + else return -1; +} + /* used to populate and activate dimms at boot time */ void dimm_scan_populated(void) { diff --git a/hw/dimm.h b/hw/dimm.h index b563e3f..0fdf59b 100644 --- a/hw/dimm.h +++ b/hw/dimm.h @@ -60,6 +60,6 @@ void dimm_activate(DimmState *slot); void dimm_deactivate(DimmState *slot); void dimm_scan_populated(void); void dimm_notify(uint32_t idx, uint32_t event); - +int dimm_set_populated(DimmState *s); #endif diff --git a/qemu-config.c b/qemu-config.c index 4abc31b..7f63186 100644 --- a/qemu-config.c +++ b/qemu-config.c @@ -650,6 +650,49 @@ static QemuOptsList qemu_dimm_opts = { { /* end of list */ } }, }; + +static QemuOptsList qemu_dimms_opts = { + .name = "dimms", + .head = QTAILQ_HEAD_INITIALIZER(qemu_dimms_opts.head), + .desc = { + { + .name = "pfx", + .type = QEMU_OPT_STRING, + .help = "prefix of ids for these dimm devices", + },{ + .name = "size", + .type = QEMU_OPT_SIZE, + .help = "memory size for these dimm", + },{ + .name = "num", + .type = QEMU_OPT_NUMBER, + .help = "number of dimm devices in this pool", + },{ + .name = "node", + .type = QEMU_OPT_NUMBER, + .help = "NUMA node number (i.e. proximity) for these dimms", + }, + { /* end of list */ } + }, +}; + +static QemuOptsList qemu_dimmspop_opts = { + .name = "dimmspop", + .head = QTAILQ_HEAD_INITIALIZER(qemu_dimmspop_opts.head), + .desc = { + { + .name = "pfx", + .type = QEMU_OPT_STRING, + .help = "pool prefix for this dimm device", + },{ + .name = "num", + .type = QEMU_OPT_SIZE, + .help = "number of dimm devices to populate", + }, + { /* end of list */ } + }, +}; + static QemuOptsList *vm_config_groups[32] = { &qemu_drive_opts, &qemu_chardev_opts, @@ -666,6 +709,8 @@ static QemuOptsList *vm_config_groups[32] = { &qemu_boot_opts, &qemu_iscsi_opts, &qemu_dimm_opts, + &qemu_dimms_opts, + &qemu_dimmspop_opts, NULL, }; diff --git a/qemu-options.hx b/qemu-options.hx index 61909f7..0a9326e 100644 --- a/qemu-options.hx +++ b/qemu-options.hx @@ -2752,3 +2752,13 @@ DEF("dimm", HAS_ARG, QEMU_OPTION_dimm, "-dimm id=dimmid,size=sz,node=nd,populated=on|off\n" "specify memory dimm device with name dimmid, size sz on node nd", QEMU_ARCH_ALL) + +DEF("dimms", HAS_ARG, QEMU_OPTION_dimms, + "-dimms pfx=id,size=sz,node=nd\n" + "specify pool of num memory dimm devices of size sz each on node nd", + QEMU_ARCH_ALL) + +DEF("dimmspop", HAS_ARG, QEMU_OPTION_dimmspop, + "-dimmspop pfx=id,num=n\n" + "populate n dimms of pool id (dimms with ids id0,...,idn-1) at system startup", + QEMU_ARCH_ALL) diff --git a/vl.c b/vl.c index efe915e..37752be 100644 --- a/vl.c +++ b/vl.c @@ -538,6 +538,65 @@ static void configure_dimm(QemuOpts *opts) nb_hp_dimms++; } +static void configure_dimms(QemuOpts *opts) +{ + const char *value, *pfx, *id; + uint64_t size, node; + int num, dimm; + char buf[32]; + + id = qemu_opts_id(opts); + value = qemu_opt_get(opts, "pfx"); + if (!value) { + fprintf(stderr, "qemu: invalid prefix for dimm pool '%s'\n", id); + exit(1); + } + pfx = value; + + size = qemu_opt_get_size(opts, "size", DEFAULT_DIMMSIZE); + num = qemu_opt_get_number(opts, "num", 1); + node = qemu_opt_get_number(opts, "node", 0); + + for (dimm = 0; dimm < num; dimm++) { + if (nb_hp_dimms == MAX_DIMMS) { + fprintf(stderr, "qemu: maximum number of DIMMs (%d) exceeded\n", + MAX_DIMMS); + exit(1); + } + sprintf(buf, "%s%d", pfx, dimm); + dimm_create(g_strdup(buf), size, node, nb_hp_dimms, false); + nb_hp_dimms++; + } +} + +/* populate dimms at startup */ +static void configure_dimmspop(QemuOpts *opts) +{ + const char *value, *pfx, *id; + int num, dimm; + char buf[32]; + + id = qemu_opts_id(opts); + value = qemu_opt_get(opts, "pfx"); + if (!value) { + fprintf(stderr, "qemu: invalid prefix for dimm pool '%s'\n", id); + exit(1); + } + pfx = value; + value = qemu_opt_get(opts, "num"); + if (!value) { + fprintf(stderr, "qemu: number not defined for dimm pool '%s'\n", pfx); + exit(1); + } + else num = atoi(value); + for (dimm = 0; dimm < num; dimm++) { + sprintf(buf, "%s%d", pfx, dimm); + if (dimm_set_populated(dimm_find_from_name(buf)) < 0) { + fprintf(stderr, "qemu: dimm %s not defined for dimm pool '%s'\n", + buf, pfx); + } + } +} static void configure_rtc(QemuOpts *opts) { const char *value; @@ -2293,7 +2352,9 @@ int main(int argc, char **argv, char **envp) int cyls, heads, secs, translation; QemuOpts *hda_opts = NULL, *opts, *machine_opts; QemuOpts *dimm_opts[MAX_DIMMS]; - int nb_dimm_opts = 0; + QemuOpts *dimms_opts[MAX_DIMMS]; + QemuOpts *dimmspop_opts[MAX_DIMMS]; + int nb_dimm_opts = 0, nb_dimms_opts = 0, nb_dimmspop_opts = 0; QemuOptsList *olist; int optind; const char *optarg; @@ -3233,6 +3294,22 @@ int main(int argc, char **argv, char **envp) } nb_dimm_opts++; break; + case QEMU_OPTION_dimms: + dimms_opts[nb_dimms_opts] = + qemu_opts_parse(qemu_find_opts("dimms"), optarg, 0); + if (!dimms_opts[nb_dimms_opts]) { + exit(1); + } + nb_dimms_opts++; + break; + case QEMU_OPTION_dimmspop: + dimmspop_opts[nb_dimmspop_opts] = + qemu_opts_parse(qemu_find_opts("dimmspop"), optarg, 0); + if (!dimmspop_opts[nb_dimmspop_opts]) { + exit(1); + } + nb_dimmspop_opts++; + break; default: os_parse_cmd_args(popt->index, optarg); } @@ -3552,6 +3629,13 @@ int main(int argc, char **argv, char **envp) for (i = 0; i < nb_dimm_opts; i++) configure_dimm(dimm_opts[i]); + + for (i = 0; i < nb_dimms_opts; i++) + configure_dimms(dimms_opts[i]); + + for (i = 0; i < nb_dimmspop_opts; i++) + configure_dimmspop(dimmspop_opts[i]); + qdev_machine_init(); machine->init(ram_size, boot_devices, -- 1.7.9

1 0

[RFC PATCH v2 19/21] Implement "info memtotal" and "query-memtotal"
by Vasilis Liaskovitis July 11, 2012

July 11, 2012

Returns total memory of guest in bytes, including hotplugged memory. Signed-off-by: Vasilis Liaskovitis <vasilis.liaskovitis(a)profitbricks.com> --- hmp-commands.hx | 2 ++ hmp.c | 7 +++++++ hmp.h | 1 + hw/dimm.c | 15 +++++++++++++++ monitor.c | 7 +++++++ qapi-schema.json | 12 ++++++++++++ qmp-commands.hx | 20 ++++++++++++++++++++ 7 files changed, 64 insertions(+), 0 deletions(-) diff --git a/hmp-commands.hx b/hmp-commands.hx index 3172cde..016062e 100644 --- a/hmp-commands.hx +++ b/hmp-commands.hx @@ -1461,6 +1461,8 @@ show qdev device model list show roms @item info memhp show memhp +@item info memtotal +show memtotal @end table ETEXI diff --git a/hmp.c b/hmp.c index ec25d9a..8f89c7d 100644 --- a/hmp.c +++ b/hmp.c @@ -1017,3 +1017,10 @@ void hmp_info_memhp(Monitor *mon) qapi_free_MemHpInfoList(info); } + +void hmp_info_memtotal(Monitor *mon) +{ + uint64_t ram_total; + ram_total = (uint64_t)qmp_query_memtotal(NULL); + monitor_printf(mon, "MemTotal: %lu \n", ram_total); +} diff --git a/hmp.h b/hmp.h index 971e7c4..d6e715e 100644 --- a/hmp.h +++ b/hmp.h @@ -65,5 +65,6 @@ void hmp_dump_guest_memory(Monitor *mon, const QDict *qdict); void hmp_netdev_add(Monitor *mon, const QDict *qdict); void hmp_netdev_del(Monitor *mon, const QDict *qdict); void hmp_info_memhp(Monitor *mon); +void hmp_info_memtotal(Monitor *mon); #endif diff --git a/hw/dimm.c b/hw/dimm.c index 6e324d3..b544173 100644 --- a/hw/dimm.c +++ b/hw/dimm.c @@ -28,6 +28,7 @@ static dimm_hotplug_fn dimm_hotplug; static dimm_hotplug_fn dimm_revert; static QTAILQ_HEAD(Dimmlist, DimmState) dimmlist; static QTAILQ_HEAD(dimm_hp_result_head, dimm_hp_result) dimm_hp_result_queue; +extern ram_addr_t ram_size; static Property dimm_properties[] = { DEFINE_PROP_END_OF_LIST() @@ -292,6 +293,20 @@ MemHpInfoList *qmp_query_memhp(Error **errp) return head; } + +int64_t qmp_query_memtotal(Error **errp) +{ + DimmState *slot; + uint64_t info = ram_size; + + QTAILQ_FOREACH(slot, &dimmlist, nextdimm) { + if (slot->populated) { + info += slot->size; + } + } + return (int64_t)info; +} + static int dimm_init(SysBusDevice *s) { DimmState *slot; diff --git a/monitor.c b/monitor.c index 4a14e26..1dd646c 100644 --- a/monitor.c +++ b/monitor.c @@ -2739,6 +2739,13 @@ static mon_cmd_t info_cmds[] = { .mhandler.info = hmp_info_memhp, }, { + .name = "memtotal", + .args_type = "", + .params = "", + .help = "show total memory size", + .mhandler.info = hmp_info_memtotal, + }, + { .name = NULL, }, }; diff --git a/qapi-schema.json b/qapi-schema.json index 049f6f9..5bbf2c0 100644 --- a/qapi-schema.json +++ b/qapi-schema.json @@ -1888,3 +1888,15 @@ # Since: 1.1.3 ## { 'command': 'query-memhp', 'returns': ['MemHpInfo'] } + +## +# @query-memtotal: +# +# Returns total memory in bytes, including hotplugged dimms +# +# Returns: a l +# +# Since: 1.2 +## +{ 'command': 'query-memtotal', 'returns': 'int' } + diff --git a/qmp-commands.hx b/qmp-commands.hx index cd1d5f0..6c71696 100644 --- a/qmp-commands.hx +++ b/qmp-commands.hx @@ -2286,3 +2286,23 @@ Example: } EQMP + + { + .name = "query-memtotal", + .args_type = "", + .mhandler.cmd_new = qmp_marshal_input_query_memtotal + }, +SQMP +query-memtotal +---------- + +Return total memory in bytes, including hotplugged dimms + +Example: + +-> { "execute": "query-memtotal" } +<- { + "return": 1073741824 + } + +EQMP -- 1.7.9

1 0