[Qemu-devel] Re: [kvm-devel] expose host CPU features to guests

[Anthony Liguori]

On Wed, 2007-09-05 at 20:45 +0300, address@hidden wrote:
> Hi,
> 
> It's a pity not to use a host CPU feature if it is available. This patch
> exposes host CPU features to guests. It allows fine-tuning the presented
> features from the command-line.
> 
> The code could use some serious clean ups, but I think it is interesting
> enough right now. I'd be happy to hear your opinion and suggestions. 
> The diff are done against qemu cvs. I tried it with kvm, but I thinkg it
> should be useful also for kqemu.

I like this idea but I have some suggestions about the general approach.
I think instead of defining another machine type, it would be better to
just have a command line option like -cpuid that took a comma separate
string of features with "all" meaning all features that the host has.

I also think it would be nicer to use cpuid() directly instead of
attempting to parse /proc/cpuinfo.

Regards,

Anthony Liguori

> Regards,
> 
> Dan.
> 
> Index: vl.c
> ===================================================================
> RCS file: /sources/qemu/qemu/vl.c,v
> retrieving revision 1.336
> diff -u -p -r1.336 vl.c
> --- vl.c      28 Aug 2007 22:21:40 -0000      1.336
> +++ vl.c      5 Sep 2007 15:48:10 -0000
> @@ -6575,12 +6575,24 @@ int qemu_register_machine(QEMUMachine *m
>      return 0;
>  }
>  
> +const char *machine_flavor = 0;
> +
>  QEMUMachine *find_machine(const char *name)
>  {
>      QEMUMachine *m;
>  
> +    int n;
> +    machine_flavor = strchr(name, ',');
> +    if (machine_flavor) {
> +     n = machine_flavor - name;
> +     machine_flavor++;
> +    } else {
> +        n = strlen(name);
> +        machine_flavor = "";
> +    }
> +
>      for(m = first_machine; m != NULL; m = m->next) {
> -        if (!strcmp(m->name, name))
> +        if (!strncmp(m->name, name, n))
>              return m;
>      }
>      return NULL;
> @@ -7343,6 +7355,7 @@ static void read_passwords(void)
>  void register_machines(void)
>  {
>  #if defined(TARGET_I386)
> +    qemu_register_machine(&host_pc_machine);
>      qemu_register_machine(&pc_machine);
>      qemu_register_machine(&isapc_machine);
>  #elif defined(TARGET_PPC)
> Index: vl.h
> ===================================================================
> RCS file: /sources/qemu/qemu/vl.h,v
> retrieving revision 1.264
> diff -u -p -r1.264 vl.h
> --- vl.h      26 Aug 2007 17:46:00 -0000      1.264
> +++ vl.h      5 Sep 2007 15:48:10 -0000
> @@ -1156,6 +1156,7 @@ void piix4_smbus_register_device(SMBusDe
>  void acpi_bios_init(void);
>  
>  /* pc.c */
> +extern QEMUMachine host_pc_machine;
>  extern QEMUMachine pc_machine;
>  extern QEMUMachine isapc_machine;
>  extern int fd_bootchk;
> Index: hw/pc.c
> ===================================================================
> RCS file: /sources/qemu/qemu/hw/pc.c,v
> retrieving revision 1.83
> diff -u -p -r1.83 pc.c
> --- hw/pc.c   26 Aug 2007 17:51:39 -0000      1.83
> +++ hw/pc.c   5 Sep 2007 15:48:10 -0000
> @@ -965,6 +965,28 @@ static void pc_init_isa(int ram_size, in
>               initrd_filename, 0);
>  }
>  
> +int use_hostlike_cpu = 0;
> +
> +static void pc_init_hostlike(ram_addr_t ram_size, int vga_ram_size, int 
> boot_device,
> +                        DisplayState *ds, const char **fd_filename, 
> +                        int snapshot, 
> +                        const char *kernel_filename, 
> +                        const char *kernel_cmdline,
> +                        const char *initrd_filename)
> +{
> +    use_hostlike_cpu = 1;
> +    pc_init1(ram_size, vga_ram_size, boot_device,
> +             ds, fd_filename, snapshot,
> +             kernel_filename, kernel_cmdline,
> +             initrd_filename, 1);
> +}
> +
> +QEMUMachine host_pc_machine = {
> +    "host",
> +    "Standard PC with host-like CPU",
> +    pc_init_hostlike,
> +};
> +
>  QEMUMachine pc_machine = {
>      "pc",
>      "Standard PC",
> Index: target-i386/cpu.h
> ===================================================================
> RCS file: /sources/qemu/qemu/target-i386/cpu.h,v
> retrieving revision 1.45
> diff -u -p -r1.45 cpu.h
> --- target-i386/cpu.h 11 Jul 2007 22:48:58 -0000      1.45
> +++ target-i386/cpu.h 5 Sep 2007 15:48:10 -0000
> @@ -267,21 +267,44 @@
>  #define CPUID_CMOV (1 << 15)
>  #define CPUID_PAT  (1 << 16)
>  #define CPUID_PSE36   (1 << 17)
> +#define CPUID_PN   (1 << 18)
>  #define CPUID_CLFLUSH (1 << 19)
> -/* ... */
> +#define CPUID_DTS (1 << 21)
> +#define CPUID_ACPI (1 << 22)
>  #define CPUID_MMX  (1 << 23)
>  #define CPUID_FXSR (1 << 24)
>  #define CPUID_SSE  (1 << 25)
>  #define CPUID_SSE2 (1 << 26)
> +#define CPUID_SS (1 << 27)
> +#define CPUID_HT (1 << 28)
> +#define CPUID_TM (1 << 29)
> +#define CPUID_IA64 (1 << 30)
> +#define CPUID_PBE (1 << 31)
>  
>  #define CPUID_EXT_SSE3     (1 << 0)
>  #define CPUID_EXT_MONITOR  (1 << 3)
> +#define CPUID_EXT_DSCPL    (1 << 4)
> +#define CPUID_EXT_VMX      (1 << 5)
> +#define CPUID_EXT_SMX      (1 << 6)
> +#define CPUID_EXT_EST      (1 << 7)
> +#define CPUID_EXT_TM2      (1 << 8)
> +#define CPUID_EXT_SSSE3    (1 << 9)
> +#define CPUID_EXT_CID      (1 << 10)
>  #define CPUID_EXT_CX16     (1 << 13)
> +#define CPUID_EXT_XTPR     (1 << 14)
> +#define CPUID_EXT_DCA      (1 << 17)
> +#define CPUID_EXT_POPCNT   (1 << 22)
>  
>  #define CPUID_EXT2_SYSCALL (1 << 11)
> +#define CPUID_EXT2_MP      (1 << 19)
>  #define CPUID_EXT2_NX      (1 << 20)
> +#define CPUID_EXT2_MMXEXT  (1 << 22)
>  #define CPUID_EXT2_FFXSR   (1 << 25)
> +#define CPUID_EXT2_PDPE1GB (1 << 26)
> +#define CPUID_EXT2_RDTSCP  (1 << 27)
>  #define CPUID_EXT2_LM      (1 << 29)
> +#define CPUID_EXT2_3DNOWEXT (1 << 30)
> +#define CPUID_EXT2_3DNOW   (1 << 31)
>  
>  #define EXCP00_DIVZ  0
>  #define EXCP01_SSTP  1
> Index: target-i386/helper2.c
> ===================================================================
> RCS file: /sources/qemu/qemu/target-i386/helper2.c,v
> retrieving revision 1.48
> diff -u -p -r1.48 helper2.c
> --- target-i386/helper2.c     31 Jul 2007 23:09:18 -0000      1.48
> +++ target-i386/helper2.c     5 Sep 2007 15:48:10 -0000
> @@ -45,6 +45,272 @@ int modify_ldt(int func, void *ptr, unsi
>  #endif
>  #endif /* USE_CODE_COPY */
>  
> +/* x86_cap_flags taken from Linux's arch/i386/kernel/cpu/proc.c */
> +static const char * const x86_cap_flags[] = {
> +     /* Intel-defined */
> +        "fpu", "vme", "de", "pse", "tsc", "msr", "pae", "mce",
> +        "cx8", "apic", NULL, "sep", "mtrr", "pge", "mca", "cmov",
> +        "pat", "pse36", "pn", "clflush", NULL, "dts", "acpi", "mmx",
> +        "fxsr", "sse", "sse2", "ss", "ht", "tm", "ia64", "pbe",
> +
> +     /* AMD-defined */
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, "syscall", NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, "mp", "nx", NULL, "mmxext", NULL,
> +     NULL, "fxsr_opt", "pdpe1gb", "rdtscp", NULL, "lm", "3dnowext", "3dnow",
> +
> +     /* Transmeta-defined */
> +     "recovery", "longrun", NULL, "lrti", NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +
> +     /* Other (Linux-defined) */
> +     "cxmmx", "k6_mtrr", "cyrix_arr", "centaur_mcr",
> +     NULL, NULL, NULL, NULL,
> +     "constant_tsc", "up", NULL, NULL, NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +
> +     /* Intel-defined (#2) */
> +     "pni", NULL, NULL, "monitor", "ds_cpl", "vmx", "smx", "est",
> +     "tm2", "ssse3", "cid", NULL, NULL, "cx16", "xtpr", NULL,
> +     NULL, NULL, "dca", NULL, NULL, NULL, NULL, "popcnt",
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +
> +     /* VIA/Cyrix/Centaur-defined */
> +     NULL, NULL, "rng", "rng_en", NULL, NULL, "ace", "ace_en",
> +     "ace2", "ace2_en", "phe", "phe_en", "pmm", "pmm_en", NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +
> +     /* AMD-defined (#2) */
> +     "lahf_lm", "cmp_legacy", "svm", "extapic", "cr8legacy", "abm",
> +     "sse4a", "misalignsse",
> +     "3dnowprefetch", "osvw", "ibs", NULL, NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +     NULL, NULL, NULL, NULL, NULL, NULL, NULL, NULL,
> +};
> +
> +void add_flagname_to_bitmaps(char *flagname, int *features, int 
> *ext_features, int *ext2_features)
> +{
> +    int i;
> +    for ( i = 0 ; i < 32 ; i++ ) 
> +        if (x86_cap_flags[i] && !strcmp (flagname, x86_cap_flags[i])) {
> +            *features |= 1 << i;
> +            return;
> +        }
> +    for ( i = 0 ; i < 32 ; i++ ) 
> +        if (x86_cap_flags[32*4+i] && !strcmp (flagname, 
> x86_cap_flags[32*4+i])) {
> +            *ext_features |= 1 << i;
> +            return;
> +        }
> +    for ( i = 0 ; i < 32 ; i++ ) 
> +        if (x86_cap_flags[32*1+i] && !strcmp (flagname, 
> x86_cap_flags[32*1+i])) {
> +            *ext2_features |= 1 << i;
> +            return;
> +        }
> +    /* Silently ignore Linux-defined features */
> +    for ( i = 0 ; i < 32 ; i++ ) 
> +        if (x86_cap_flags[32*3+i] && !strcmp (flagname, 
> x86_cap_flags[32*3+i])) {
> +            return;
> +        }
> +    fprintf(stderr, "CPU feature %s not found\n", flagname);
> +}
> +
> +static void apply_machine_flavor(CPUX86State *env)
> +{
> +    int plus_features = 0, plus_ext_features = 0, plus_ext2_features = 0;
> +    int minus_features = 0, minus_ext_features = 0, minus_ext2_features = 0;
> +    int family = -1, model = -1, stepping = -1;
> +
> +    extern char *machine_flavor;
> +    char *s = strdup(machine_flavor);
> +    char *featurestr = strtok(s, ",");
> +
> +    while (featurestr) {
> +      char *val;
> +      if (featurestr[0] == '+') {
> +         add_flagname_to_bitmaps(featurestr + 1, &plus_features, 
> &plus_ext_features, &plus_ext2_features);
> +      }
> +      else if (featurestr[0] == '-') {
> +         add_flagname_to_bitmaps(featurestr + 1, &minus_features, 
> &minus_ext_features, &minus_ext2_features);
> +      } 
> +      else if ((val = strchr(featurestr, '='))) {
> +         *val = 0; val++;
> +         if (!strcmp(featurestr, "family")) {
> +            family = atoi(val);
> +            if (family <= 0) {
> +               fprintf(stderr, "bad numerical value %s\n", val);
> +               exit(1);
> +            }
> +            env->cpuid_version &= 0xFF;
> +            env->cpuid_version |= (family << 8);
> +         } else if (!strcmp(featurestr, "model")) {
> +            model = atoi(val);
> +            if (model <= 0) {
> +               fprintf(stderr, "bad numerical value %s\n", val);
> +               exit(1);
> +            }
> +            env->cpuid_version &= 0xFFFFFF0F;
> +            env->cpuid_version |= (model << 4);
> +         } else if (!strcmp(featurestr, "stepping")) {
> +            stepping = atoi(val);
> +            if (stepping <= 0) {
> +               fprintf(stderr, "bad numerical value %s\n", val);
> +               exit(1);
> +            }
> +            env->cpuid_version &= 0xFFFFFFF0;
> +            env->cpuid_version |= stepping;
> +         } else {
> +            fprintf(stderr, "unknown feature %s\n", featurestr);
> +            exit(1);
> +         }
> +      } else {
> +         fprintf(stderr, "feature string `%s' not in format 
> (+feature|-feature|feature=xyz)\n", featurestr);
> +         exit(1);
> +      }
> +      featurestr = strtok(NULL, ",");
> +    }
> +    free(s);
> +    env->cpuid_features |= plus_features;
> +    env->cpuid_ext_features |= plus_ext_features;
> +    env->cpuid_ext2_features |= plus_ext2_features;
> +    env->cpuid_features &= ~minus_features;
> +    env->cpuid_ext_features &= ~minus_ext_features;
> +    env->cpuid_ext2_features &= ~minus_ext2_features;
> +}
> +
> +static int read_one_cpu_info(CPUX86State *env, FILE *f)
> +{
> +    char line[300], *val;
> +    char vendor_id[13];
> +    int family = -1, model = -1, stepping = -1;
> +    int features = 0, ext_features = 0, ext2_features = 0;
> +
> +    while (fgets(line, sizeof(line), f)) {
> +        if (!strtok(line, "\t") || line[0]=='\n') {
> +            break;
> +        }
> +        val = strtok(NULL, "\t:");
> +        if (!val)
> +            val = "";
> +        if (*val==' ')
> +            val++;
> +
> +        if (!strcmp(line,"vendor_id")) {
> +            strncpy(vendor_id, val, sizeof(vendor_id) - 1);
> +            vendor_id[sizeof(vendor_id)-1] = 0;
> +        }
> +        if (!strcmp(line,"cpu family")) {
> +            family = atoi(val);
> +        }
> +        if (!strcmp(line,"model")) {
> +            model = atoi(val);
> +        }
> +        if (!strcmp(line,"stepping")) {
> +            stepping = atoi(val);
> +        }
> +        if (!strcmp(line,"flags")) {
> +            char *flagname;
> +            struct flagstorage {
> +                char *name;
> +                int  bit;
> +                int  *container;
> +            };
> +
> +            flagname = strtok(val, " ");
> +            while (flagname) {
> +                add_flagname_to_bitmaps(flagname, &features, &ext_features, 
> &ext2_features);
> +                flagname = strtok(NULL, " \n");
> +            }
> +            features &= ~CPUID_ACPI; /* acpi causes guest kernel panic on 
> boot time, in cpu_identify */
> +            ext_features &= ~CPUID_EXT_VMX; /* KVM is not recursive */
> +        }
> +    }
> +    if (family==-1 || model==-1 || stepping==-1)
> +        return -1;
> +    env->cpuid_level = 2;
> +    env->cpuid_vendor1 = *(uint32_t *)&vendor_id[0];
> +    env->cpuid_vendor2 = *(uint32_t *)&vendor_id[4];
> +    env->cpuid_vendor3 = *(uint32_t *)&vendor_id[8];
> +    env->cpuid_version = (family << 8) | (model << 4) | stepping;
> +    env->cpuid_features = features;
> +    env->cpuid_ext_features = ext_features;
> +    env->cpuid_ext2_features = ext2_features;
> +    env->pat = 0x0007040600070406ULL;
> +    /* TODO expose real cpuid extended level */
> +#ifdef TARGET_X86_64
> +    env->cpuid_xlevel = 0x80000008;
> +#else
> +    env->cpuid_xlevel = 0;
> +#endif
> +    return 0;
> +}
> +
> +static int set_guest_cpu_hostlike(CPUX86State *env) {
> +    FILE *f;
> +    f = fopen("/proc/cpuinfo", "r");
> +    if (!f) {
> +      fprintf(stderr, "falling back to standard guest cpu\n");
> +      return -1;
> +    }
> +    read_one_cpu_info(env, f);
> +    fclose(f);
> +    return 0;
> +}
> +
> +static int set_guest_cpu_standard(CPUX86State *env)
> +{
> +    int family, model, stepping;
> +#ifdef TARGET_X86_64
> +    env->cpuid_vendor1 = 0x68747541; /* "Auth" */
> +    env->cpuid_vendor2 = 0x69746e65; /* "enti" */
> +    env->cpuid_vendor3 = 0x444d4163; /* "cAMD" */
> +    family = 6;
> +    model = 2;
> +    stepping = 3;
> +#else
> +    env->cpuid_vendor1 = 0x756e6547; /* "Genu" */
> +    env->cpuid_vendor2 = 0x49656e69; /* "ineI" */
> +    env->cpuid_vendor3 = 0x6c65746e; /* "ntel" */
> +#if 0
> +    /* pentium 75-200 */
> +    family = 5;
> +    model = 2;
> +    stepping = 11;
> +#else
> +    /* pentium pro */
> +    family = 6;
> +    model = 3;
> +    stepping = 3;
> +#endif
> +#endif
> +    env->cpuid_level = 2;
> +    env->cpuid_version = (family << 8) | (model << 4) | stepping;
> +    env->cpuid_features = (CPUID_FP87 | CPUID_DE | CPUID_PSE |
> +            CPUID_TSC | CPUID_MSR | CPUID_MCE |
> +            CPUID_CX8 | CPUID_PGE | CPUID_CMOV |
> +            CPUID_PAT);
> +    env->pat = 0x0007040600070406ULL;
> +    env->cpuid_ext_features = CPUID_EXT_SSE3;
> +    env->cpuid_features |= CPUID_FXSR | CPUID_MMX | CPUID_SSE | CPUID_SSE2 | 
> CPUID_PAE | CPUID_SEP;
> +    env->cpuid_features |= CPUID_APIC;
> +    env->cpuid_xlevel = 0;
> +#ifdef TARGET_X86_64
> +    /* currently not enabled for std i386 because not fully tested */
> +    env->cpuid_ext2_features = (env->cpuid_features & 0x0183F3FF);
> +    env->cpuid_ext2_features |= CPUID_EXT2_LM | CPUID_EXT2_SYSCALL | 
> CPUID_EXT2_NX;
> +    env->cpuid_xlevel = 0x80000008;
> +
> +    /* these features are needed for Win64 and aren't fully implemented */
> +    env->cpuid_features |= CPUID_MTRR | CPUID_CLFLUSH | CPUID_MCA;
> +    /* this feature is needed for Solaris and isn't fully implemented */
> +    env->cpuid_features |= CPUID_PSE36;
> +#endif
> +    return 0;
> +}
> +
>  CPUX86State *cpu_x86_init(void)
>  {
>      CPUX86State *env;
> @@ -80,64 +346,23 @@ CPUX86State *cpu_x86_init(void)
>      }
>  #endif
>      {
> -        int family, model, stepping;
> -#ifdef TARGET_X86_64
> -        env->cpuid_vendor1 = 0x68747541; /* "Auth" */
> -        env->cpuid_vendor2 = 0x69746e65; /* "enti" */
> -        env->cpuid_vendor3 = 0x444d4163; /* "cAMD" */
> -        family = 6;
> -        model = 2;
> -        stepping = 3;
> -#else
> -        env->cpuid_vendor1 = 0x756e6547; /* "Genu" */
> -        env->cpuid_vendor2 = 0x49656e69; /* "ineI" */
> -        env->cpuid_vendor3 = 0x6c65746e; /* "ntel" */
> -#if 0
> -        /* pentium 75-200 */
> -        family = 5;
> -        model = 2;
> -        stepping = 11;
> -#else
> -        /* pentium pro */
> -        family = 6;
> -        model = 3;
> -        stepping = 3;
> -#endif
> -#endif
> -        env->cpuid_level = 2;
> -        env->cpuid_version = (family << 8) | (model << 4) | stepping;
> -        env->cpuid_features = (CPUID_FP87 | CPUID_DE | CPUID_PSE |
> -                               CPUID_TSC | CPUID_MSR | CPUID_MCE |
> -                               CPUID_CX8 | CPUID_PGE | CPUID_CMOV |
> -                               CPUID_PAT);
> -        env->pat = 0x0007040600070406ULL;
> -        env->cpuid_ext_features = CPUID_EXT_SSE3;
> -        env->cpuid_features |= CPUID_FXSR | CPUID_MMX | CPUID_SSE | 
> CPUID_SSE2 | CPUID_PAE | CPUID_SEP;
> -        env->cpuid_features |= CPUID_APIC;
> -        env->cpuid_xlevel = 0;
> -        {
> -            const char *model_id = "QEMU Virtual CPU version " QEMU_VERSION;
> -            int c, len, i;
> -            len = strlen(model_id);
> -            for(i = 0; i < 48; i++) {
> -                if (i >= len)
> -                    c = '\0';
> -                else
> -                    c = model_id[i];
> -                env->cpuid_model[i >> 2] |= c << (8 * (i & 3));
> -            }
> +        extern int use_hostlike_cpu;
> +        if (!use_hostlike_cpu || set_guest_cpu_hostlike(env))
> +            set_guest_cpu_standard(env);
> +        apply_machine_flavor(env);
> +    }
> +    {
> +        const char *model_id = "QEMU Virtual CPU version " QEMU_VERSION;
> +        int c, len, i;
> +
> +        len = strlen(model_id);
> +        for(i = 0; i < 48; i++) {
> +            if (i >= len)
> +                c = '\0';
> +            else
> +                c = model_id[i];
> +            env->cpuid_model[i >> 2] |= c << (8 * (i & 3));
>          }
> -#ifdef TARGET_X86_64
> -        /* currently not enabled for std i386 because not fully tested */
> -        env->cpuid_ext2_features = (env->cpuid_features & 0x0183F3FF);
> -        env->cpuid_ext2_features |= CPUID_EXT2_LM | CPUID_EXT2_SYSCALL | 
> CPUID_EXT2_NX;
> -        env->cpuid_xlevel = 0x80000008;
> -
> -        /* these features are needed for Win64 and aren't fully implemented 
> */
> -        env->cpuid_features |= CPUID_MTRR | CPUID_CLFLUSH | CPUID_MCA;
> -        /* this feature is needed for Solaris and isn't fully implemented */
> -        env->cpuid_features |= CPUID_PSE36;
> -#endif
>      }
>      cpu_reset(env);
>  #ifdef USE_KQEMU
> 
> -------------------------------------------------------------------------
> This SF.net email is sponsored by: Splunk Inc.
> Still grepping through log files to find problems?  Stop.
> Now Search log events and configuration files using AJAX and a browser.
> Download your FREE copy of Splunk now >>  http://get.splunk.com/
> _______________________________________________
> kvm-devel mailing list
> address@hidden
> https://lists.sourceforge.net/lists/listinfo/kvm-devel