From: Xiao Guangrong <guangrong.xiao@linux.intel.com> To: Stefan Hajnoczi <stefanha@gmail.com> Cc: pbonzini@redhat.com, imammedo@redhat.com, gleb@kernel.org, mtosatti@redhat.com, stefanha@redhat.com, mst@redhat.com, rth@twiddle.net, ehabkost@redhat.com, kvm@vger.kernel.org, qemu-devel@nongnu.org Subject: Re: [PATCH v2 08/18] nvdimm: init backend memory mapping and config data area Date: Wed, 26 Aug 2015 18:40:26 +0800 [thread overview] Message-ID: <55DD979A.70804@linux.intel.com> (raw) In-Reply-To: <20150825160353.GD8344@stefanha-thinkpad.redhat.com> On 08/26/2015 12:03 AM, Stefan Hajnoczi wrote: > On Fri, Aug 14, 2015 at 10:52:01PM +0800, Xiao Guangrong wrote: >> The parameter @file is used as backed memory for NVDIMM which is >> divided into two parts if @dataconfig is true: > > s/dataconfig/configdata/ Stupid typo, sorry. > >> @@ -76,13 +109,87 @@ static void pc_nvdimm_init(Object *obj) >> set_configdata, NULL); >> } >> >> +static uint64_t get_file_size(int fd) >> +{ >> + struct stat stat_buf; >> + uint64_t size; >> + >> + if (fstat(fd, &stat_buf) < 0) { >> + return 0; >> + } >> + >> + if (S_ISREG(stat_buf.st_mode)) { >> + return stat_buf.st_size; >> + } >> + >> + if (S_ISBLK(stat_buf.st_mode) && !ioctl(fd, BLKGETSIZE64, &size)) { >> + return size; >> + } > > #ifdef __linux__ for ioctl(fd, BLKGETSIZE64, &size)? > > There is nothing Linux-specific about emulating NVDIMMs so this code > should compile on all platforms. Right. I have no idea for how block devices work on other platforms so I will only allow linux to directly use bock device file in the next version. > >> + >> + return 0; >> +} >> + >> static void pc_nvdimm_realize(DeviceState *dev, Error **errp) >> { >> PCNVDIMMDevice *nvdimm = PC_NVDIMM(dev); >> + char name[512]; >> + void *buf; >> + ram_addr_t addr; >> + uint64_t size, nvdimm_size, config_size = MIN_CONFIG_DATA_SIZE; >> + int fd; >> >> if (!nvdimm->file) { >> error_setg(errp, "file property is not set"); >> } > > Missing return here. Will fix. > >> + >> + fd = open(nvdimm->file, O_RDWR); > > Does it make sense to support read-only NVDIMMs? > > It could be handy for sharing a read-only file between unprivileged > guests. The permissions on the file would only allow read, not write. Make sense. Currently these patchset just implements "shared" mode so that write permission is required, however, please see below: > >> + if (fd < 0) { >> + error_setg(errp, "can not open %s", nvdimm->file); > > s/can not/cannot/ > >> + return; >> + } >> + >> + size = get_file_size(fd); >> + buf = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0); > > I guess the user will want to choose between MAP_SHARED and MAP_PRIVATE. > This can be added in the future. Good idea, it will allow guest to write data but discards its content after it exits. Will implement O_RDONLY + MAP_PRIVATE in the near future. > >> + if (buf == MAP_FAILED) { >> + error_setg(errp, "can not do mmap on %s", nvdimm->file); >> + goto do_close; >> + } >> + >> + nvdimm->config_data_size = config_size; >> + if (nvdimm->configdata) { >> + /* reserve MIN_CONFIGDATA_AREA_SIZE for configue data. */ >> + nvdimm_size = size - config_size; >> + nvdimm->config_data_addr = buf + nvdimm_size; >> + } else { >> + nvdimm_size = size; >> + nvdimm->config_data_addr = NULL; >> + } >> + >> + if ((int64_t)nvdimm_size <= 0) { > > The error cases can be detected before mmap(2). That avoids the int64_t > cast and also avoids nvdimm_size underflow and the bogus > nvdimm->config_data_addr calculation above. Okay. > > size = get_file_size(fd); > if (size == 0) { > error_setg(errp, "empty file or unable to get file size"); > goto do_close; > } else if (nvdimm->configdata && size < config_size) {{ > error_setg(errp, "file size is too small to store NVDIMM" > " configure data"); > goto do_close; > } > >> + error_setg(errp, "file size is too small to store NVDIMM" >> + " configure data"); >> + goto do_unmap; >> + } >> + >> + addr = reserved_range_push(nvdimm_size); >> + if (!addr) { >> + error_setg(errp, "do not have enough space for size %#lx.\n", size); > > error_setg() messages must not have a newline at the end. > > Please use "%#" PRIx64 instead of "%#lx" so compilation works on 32-bit > hosts where sizeof(long) == 4. Good catch. > >> + goto do_unmap; >> + } >> + >> + nvdimm->device_index = new_device_index(); >> + sprintf(name, "NVDIMM-%d", nvdimm->device_index); >> + memory_region_init_ram_ptr(&nvdimm->mr, OBJECT(dev), name, nvdimm_size, >> + buf); > > How is the autogenerated name used? > > Why not just use "pc-nvdimm.memory"? Ah. Just for debug proposal :) and i am not sure if a name used for multiple MRs (MemoryRegion) is a good idea. > >> + vmstate_register_ram(&nvdimm->mr, DEVICE(dev)); >> + memory_region_add_subregion(get_system_memory(), addr, &nvdimm->mr); >> + >> + return; > > fd is leaked. Will fix.
WARNING: multiple messages have this Message-ID (diff)
From: Xiao Guangrong <guangrong.xiao@linux.intel.com> To: Stefan Hajnoczi <stefanha@gmail.com> Cc: ehabkost@redhat.com, kvm@vger.kernel.org, mst@redhat.com, gleb@kernel.org, mtosatti@redhat.com, qemu-devel@nongnu.org, stefanha@redhat.com, imammedo@redhat.com, pbonzini@redhat.com, rth@twiddle.net Subject: Re: [Qemu-devel] [PATCH v2 08/18] nvdimm: init backend memory mapping and config data area Date: Wed, 26 Aug 2015 18:40:26 +0800 [thread overview] Message-ID: <55DD979A.70804@linux.intel.com> (raw) In-Reply-To: <20150825160353.GD8344@stefanha-thinkpad.redhat.com> On 08/26/2015 12:03 AM, Stefan Hajnoczi wrote: > On Fri, Aug 14, 2015 at 10:52:01PM +0800, Xiao Guangrong wrote: >> The parameter @file is used as backed memory for NVDIMM which is >> divided into two parts if @dataconfig is true: > > s/dataconfig/configdata/ Stupid typo, sorry. > >> @@ -76,13 +109,87 @@ static void pc_nvdimm_init(Object *obj) >> set_configdata, NULL); >> } >> >> +static uint64_t get_file_size(int fd) >> +{ >> + struct stat stat_buf; >> + uint64_t size; >> + >> + if (fstat(fd, &stat_buf) < 0) { >> + return 0; >> + } >> + >> + if (S_ISREG(stat_buf.st_mode)) { >> + return stat_buf.st_size; >> + } >> + >> + if (S_ISBLK(stat_buf.st_mode) && !ioctl(fd, BLKGETSIZE64, &size)) { >> + return size; >> + } > > #ifdef __linux__ for ioctl(fd, BLKGETSIZE64, &size)? > > There is nothing Linux-specific about emulating NVDIMMs so this code > should compile on all platforms. Right. I have no idea for how block devices work on other platforms so I will only allow linux to directly use bock device file in the next version. > >> + >> + return 0; >> +} >> + >> static void pc_nvdimm_realize(DeviceState *dev, Error **errp) >> { >> PCNVDIMMDevice *nvdimm = PC_NVDIMM(dev); >> + char name[512]; >> + void *buf; >> + ram_addr_t addr; >> + uint64_t size, nvdimm_size, config_size = MIN_CONFIG_DATA_SIZE; >> + int fd; >> >> if (!nvdimm->file) { >> error_setg(errp, "file property is not set"); >> } > > Missing return here. Will fix. > >> + >> + fd = open(nvdimm->file, O_RDWR); > > Does it make sense to support read-only NVDIMMs? > > It could be handy for sharing a read-only file between unprivileged > guests. The permissions on the file would only allow read, not write. Make sense. Currently these patchset just implements "shared" mode so that write permission is required, however, please see below: > >> + if (fd < 0) { >> + error_setg(errp, "can not open %s", nvdimm->file); > > s/can not/cannot/ > >> + return; >> + } >> + >> + size = get_file_size(fd); >> + buf = mmap(NULL, size, PROT_READ | PROT_WRITE, MAP_SHARED, fd, 0); > > I guess the user will want to choose between MAP_SHARED and MAP_PRIVATE. > This can be added in the future. Good idea, it will allow guest to write data but discards its content after it exits. Will implement O_RDONLY + MAP_PRIVATE in the near future. > >> + if (buf == MAP_FAILED) { >> + error_setg(errp, "can not do mmap on %s", nvdimm->file); >> + goto do_close; >> + } >> + >> + nvdimm->config_data_size = config_size; >> + if (nvdimm->configdata) { >> + /* reserve MIN_CONFIGDATA_AREA_SIZE for configue data. */ >> + nvdimm_size = size - config_size; >> + nvdimm->config_data_addr = buf + nvdimm_size; >> + } else { >> + nvdimm_size = size; >> + nvdimm->config_data_addr = NULL; >> + } >> + >> + if ((int64_t)nvdimm_size <= 0) { > > The error cases can be detected before mmap(2). That avoids the int64_t > cast and also avoids nvdimm_size underflow and the bogus > nvdimm->config_data_addr calculation above. Okay. > > size = get_file_size(fd); > if (size == 0) { > error_setg(errp, "empty file or unable to get file size"); > goto do_close; > } else if (nvdimm->configdata && size < config_size) {{ > error_setg(errp, "file size is too small to store NVDIMM" > " configure data"); > goto do_close; > } > >> + error_setg(errp, "file size is too small to store NVDIMM" >> + " configure data"); >> + goto do_unmap; >> + } >> + >> + addr = reserved_range_push(nvdimm_size); >> + if (!addr) { >> + error_setg(errp, "do not have enough space for size %#lx.\n", size); > > error_setg() messages must not have a newline at the end. > > Please use "%#" PRIx64 instead of "%#lx" so compilation works on 32-bit > hosts where sizeof(long) == 4. Good catch. > >> + goto do_unmap; >> + } >> + >> + nvdimm->device_index = new_device_index(); >> + sprintf(name, "NVDIMM-%d", nvdimm->device_index); >> + memory_region_init_ram_ptr(&nvdimm->mr, OBJECT(dev), name, nvdimm_size, >> + buf); > > How is the autogenerated name used? > > Why not just use "pc-nvdimm.memory"? Ah. Just for debug proposal :) and i am not sure if a name used for multiple MRs (MemoryRegion) is a good idea. > >> + vmstate_register_ram(&nvdimm->mr, DEVICE(dev)); >> + memory_region_add_subregion(get_system_memory(), addr, &nvdimm->mr); >> + >> + return; > > fd is leaked. Will fix.
next prev parent reply other threads:[~2015-08-26 10:46 UTC|newest] Thread overview: 164+ messages / expand[flat|nested] mbox.gz Atom feed top 2015-08-14 14:51 [PATCH v2 00/18] implement vNVDIMM Xiao Guangrong 2015-08-14 14:51 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:51 ` [PATCH v2 01/18] acpi: allow aml_operation_region() working on 64 bit offset Xiao Guangrong 2015-08-14 14:51 ` [Qemu-devel] " Xiao Guangrong 2015-09-02 8:05 ` Igor Mammedov 2015-09-02 8:05 ` [Qemu-devel] " Igor Mammedov 2015-08-14 14:51 ` [PATCH v2 02/18] i386/acpi-build: allow SSDT to operate on 64 bit Xiao Guangrong 2015-08-14 14:51 ` [Qemu-devel] " Xiao Guangrong 2015-09-02 10:06 ` Igor Mammedov 2015-09-02 10:06 ` [Qemu-devel] " Igor Mammedov 2015-09-02 10:43 ` Xiao Guangrong 2015-09-02 11:42 ` Igor Mammedov 2015-09-06 7:01 ` Xiao Guangrong 2015-09-02 12:05 ` Michael S. Tsirkin 2015-09-02 12:05 ` [Qemu-devel] " Michael S. Tsirkin 2015-08-14 14:51 ` [PATCH v2 03/18] acpi: add aml_derefof Xiao Guangrong 2015-08-14 14:51 ` [Qemu-devel] " Xiao Guangrong 2015-09-02 10:16 ` Igor Mammedov 2015-09-02 10:16 ` [Qemu-devel] " Igor Mammedov 2015-09-02 10:38 ` Xiao Guangrong 2015-09-02 10:38 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:51 ` [PATCH v2 04/18] acpi: add aml_sizeof Xiao Guangrong 2015-08-14 14:51 ` [Qemu-devel] " Xiao Guangrong 2015-09-02 10:18 ` Igor Mammedov 2015-09-02 10:18 ` [Qemu-devel] " Igor Mammedov 2015-09-02 10:39 ` Xiao Guangrong 2015-09-02 10:39 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:51 ` [PATCH v2 05/18] acpi: add aml_create_field Xiao Guangrong 2015-08-14 14:51 ` [Qemu-devel] " Xiao Guangrong 2015-09-02 11:10 ` Igor Mammedov 2015-09-02 11:10 ` [Qemu-devel] " Igor Mammedov 2015-09-06 5:32 ` Xiao Guangrong 2015-09-06 5:32 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:51 ` [PATCH v2 06/18] pc: implement NVDIMM device abstract Xiao Guangrong 2015-08-14 14:51 ` [Qemu-devel] " Xiao Guangrong 2015-08-25 14:57 ` Stefan Hajnoczi 2015-08-25 14:57 ` Stefan Hajnoczi 2015-08-26 9:37 ` Xiao Guangrong 2015-08-26 9:37 ` Xiao Guangrong 2015-09-02 9:58 ` Igor Mammedov 2015-09-02 9:58 ` Igor Mammedov 2015-09-02 10:36 ` Xiao Guangrong 2015-09-02 10:36 ` Xiao Guangrong 2015-09-02 11:31 ` Igor Mammedov 2015-09-02 11:31 ` Igor Mammedov 2015-09-06 6:07 ` Xiao Guangrong 2015-09-06 6:07 ` Xiao Guangrong 2015-09-07 13:40 ` Igor Mammedov 2015-09-07 13:40 ` Igor Mammedov 2015-09-08 14:03 ` Xiao Guangrong 2015-09-10 9:47 ` Igor Mammedov 2015-08-14 14:52 ` [PATCH v2 07/18] nvdimm: reserve address range for NVDIMM Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-25 15:12 ` Stefan Hajnoczi 2015-08-25 15:12 ` [Qemu-devel] " Stefan Hajnoczi 2015-08-26 9:39 ` Xiao Guangrong 2015-08-26 9:40 ` Xiao Guangrong 2015-08-26 9:40 ` Xiao Guangrong 2015-08-25 15:39 ` Stefan Hajnoczi 2015-08-25 15:39 ` [Qemu-devel] " Stefan Hajnoczi 2015-08-28 17:25 ` Eduardo Habkost 2015-08-28 17:25 ` [Qemu-devel] " Eduardo Habkost 2015-08-31 7:01 ` Xiao Guangrong 2015-08-31 7:01 ` [Qemu-devel] " Xiao Guangrong 2015-09-04 12:02 ` Igor Mammedov 2015-09-04 12:02 ` [Qemu-devel] " Igor Mammedov 2015-09-06 7:22 ` Xiao Guangrong 2015-09-06 7:22 ` Xiao Guangrong 2015-08-14 14:52 ` [PATCH v2 08/18] nvdimm: init backend memory mapping and config data area Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-25 16:03 ` Stefan Hajnoczi 2015-08-25 16:03 ` [Qemu-devel] " Stefan Hajnoczi 2015-08-26 10:40 ` Xiao Guangrong [this message] 2015-08-26 10:40 ` Xiao Guangrong 2015-08-28 11:58 ` Stefan Hajnoczi 2015-08-28 11:58 ` [Qemu-devel] " Stefan Hajnoczi 2015-08-31 6:23 ` Xiao Guangrong 2015-08-31 6:23 ` [Qemu-devel] " Xiao Guangrong 2015-09-01 9:14 ` Stefan Hajnoczi 2015-09-01 9:14 ` [Qemu-devel] " Stefan Hajnoczi 2015-09-15 16:10 ` Paolo Bonzini 2015-09-15 16:10 ` [Qemu-devel] " Paolo Bonzini 2015-09-17 8:39 ` Xiao Guangrong 2015-09-17 8:39 ` [Qemu-devel] " Xiao Guangrong 2015-09-17 9:04 ` Igor Mammedov 2015-09-17 9:04 ` [Qemu-devel] " Igor Mammedov 2015-09-17 9:14 ` Xiao Guangrong 2015-09-17 9:14 ` [Qemu-devel] " Xiao Guangrong 2015-09-17 9:34 ` Paolo Bonzini 2015-09-17 9:34 ` [Qemu-devel] " Paolo Bonzini 2015-09-17 12:43 ` Xiao Guangrong 2015-09-15 16:07 ` Paolo Bonzini 2015-09-15 16:07 ` [Qemu-devel] " Paolo Bonzini 2015-09-17 8:23 ` Xiao Guangrong 2015-09-17 8:23 ` [Qemu-devel] " Xiao Guangrong 2015-09-15 16:06 ` Paolo Bonzini 2015-09-15 16:06 ` [Qemu-devel] " Paolo Bonzini 2015-09-17 8:21 ` Xiao Guangrong 2015-09-17 8:21 ` [Qemu-devel] " Xiao Guangrong 2015-09-07 14:11 ` Igor Mammedov 2015-09-07 14:11 ` Igor Mammedov 2015-09-08 13:38 ` Xiao Guangrong 2015-09-10 10:35 ` Igor Mammedov 2015-09-15 16:11 ` Paolo Bonzini 2015-08-14 14:52 ` [PATCH v2 09/18] nvdimm: build ACPI NFIT table Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-09-15 16:12 ` Paolo Bonzini 2015-09-15 16:12 ` [Qemu-devel] " Paolo Bonzini 2015-09-15 17:35 ` Igor Mammedov 2015-09-15 17:35 ` [Qemu-devel] " Igor Mammedov 2015-08-14 14:52 ` [PATCH v2 10/18] nvdimm: init the address region used by DSM method Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-25 16:11 ` Stefan Hajnoczi 2015-08-25 16:11 ` [Qemu-devel] " Stefan Hajnoczi 2015-08-26 10:41 ` Xiao Guangrong 2015-08-26 10:41 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:52 ` [PATCH v2 11/18] nvdimm: build ACPI nvdimm devices Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:52 ` [PATCH v2 12/18] nvdimm: save arg3 for NVDIMM device _DSM method Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:52 ` [PATCH v2 13/18] nvdimm: build namespace config data Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-25 16:16 ` Stefan Hajnoczi 2015-08-25 16:16 ` Stefan Hajnoczi 2015-08-26 10:42 ` Xiao Guangrong 2015-08-26 10:42 ` Xiao Guangrong 2015-08-28 11:59 ` Stefan Hajnoczi 2015-08-28 11:59 ` Stefan Hajnoczi 2015-08-31 6:25 ` Xiao Guangrong 2015-08-31 6:25 ` Xiao Guangrong 2015-08-14 14:52 ` [PATCH v2 14/18] nvdimm: support NFIT_CMD_IMPLEMENTED function Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-25 16:23 ` Stefan Hajnoczi 2015-08-25 16:23 ` [Qemu-devel] " Stefan Hajnoczi 2015-08-26 10:46 ` Xiao Guangrong 2015-08-26 10:46 ` [Qemu-devel] " Xiao Guangrong 2015-08-28 12:01 ` Stefan Hajnoczi 2015-08-28 12:01 ` [Qemu-devel] " Stefan Hajnoczi 2015-08-31 6:51 ` Xiao Guangrong 2015-08-31 6:51 ` Xiao Guangrong 2015-09-01 9:16 ` Stefan Hajnoczi 2015-09-01 9:16 ` Stefan Hajnoczi 2015-08-14 14:52 ` [PATCH v2 15/18] nvdimm: support NFIT_CMD_GET_CONFIG_SIZE function Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-25 16:24 ` Stefan Hajnoczi 2015-08-25 16:24 ` [Qemu-devel] " Stefan Hajnoczi 2015-08-26 10:47 ` Xiao Guangrong 2015-08-26 10:47 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:52 ` [PATCH v2 16/18] nvdimm: support NFIT_CMD_GET_CONFIG_DATA Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:52 ` [PATCH v2 17/18] nvdimm: support NFIT_CMD_SET_CONFIG_DATA Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-14 14:52 ` [PATCH v2 18/18] nvdimm: add maintain info Xiao Guangrong 2015-08-14 14:52 ` [Qemu-devel] " Xiao Guangrong 2015-08-25 16:26 ` [Qemu-devel] [PATCH v2 00/18] implement vNVDIMM Stefan Hajnoczi 2015-08-25 16:26 ` Stefan Hajnoczi 2015-08-26 10:49 ` Xiao Guangrong 2015-08-26 10:49 ` Xiao Guangrong 2015-10-07 14:02 ` Stefan Hajnoczi 2015-10-07 14:02 ` Stefan Hajnoczi 2015-10-07 14:43 ` Xiao Guangrong 2015-10-07 14:43 ` Xiao Guangrong 2015-10-09 10:38 ` Stefan Hajnoczi 2015-10-09 10:38 ` Stefan Hajnoczi
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=55DD979A.70804@linux.intel.com \ --to=guangrong.xiao@linux.intel.com \ --cc=ehabkost@redhat.com \ --cc=gleb@kernel.org \ --cc=imammedo@redhat.com \ --cc=kvm@vger.kernel.org \ --cc=mst@redhat.com \ --cc=mtosatti@redhat.com \ --cc=pbonzini@redhat.com \ --cc=qemu-devel@nongnu.org \ --cc=rth@twiddle.net \ --cc=stefanha@gmail.com \ --cc=stefanha@redhat.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.