Re: [Qemu-devel] expose host CPU features to guests: Take 2

["Natalia Portillo" <claunia@claunia.com>]

I don't see in what is it useful without KVM/KQEMU.

And even with them there are some instructions that can't be accesible  
without KQEMU/KVM prepared for them.

And, the -cpu option, should be enabled in x86 and x86_64 to  
enable/disable emulation of instructions (and them cpuid adjusted to  
indicate them).

Regards

On Mon, 10 Sep 2007 08:40:05 +0100, Dan Kenigsberg <danken@qumranet.com>  
wrote:

> As with Take 1 of this patch, its purpose is to expose host CPU features  
> to
> guests. It proved rather helpful to KVM in various benchmarks, and it  
> should
> similarly speed kqemu up. Note that it does not extend the set of  
> emulated
> opcodes, only exposes what's supported by the host CPU.
>
> I changed the patch according to the comments on these mailing lists.  
> Thanks
> for your suggestions. Changes since "take 1" are:
> - Use the new -cpu option to specify cpu features.
> - Read host cpuid directly, instead of parsing /proc/cpuinfo. This  
> approach is
>   less Linux-centric and should work with other OSs too. It means  
> however, that
>   the cpuid is not checked for old CPU bugs. In the future, qemu should
>   negotiate the actually-supported features with the virutalizing kernel
>   module.
>
> Index: hw/pc.c
> ===================================================================
> RCS file: /sources/qemu/qemu/hw/pc.c,v
> retrieving revision 1.83
> diff -u -r1.83 pc.c
> --- hw/pc.c	26 Aug 2007 17:51:39 -0000	1.83
> +++ hw/pc.c	10 Sep 2007 06:54:00 -0000
> @@ -666,7 +666,7 @@
>                       DisplayState *ds, const char **fd_filename, int  
> snapshot,
>                       const char *kernel_filename, const char  
> *kernel_cmdline,
>                       const char *initrd_filename,
> -                     int pci_enabled)
> +                     int pci_enabled, char *cpu_model)
>  {
>      char buf[1024];
>      int ret, linux_boot, i;
> @@ -683,7 +683,9 @@
>     /* init CPUs */
>      for(i = 0; i < smp_cpus; i++) {
> +        void apply_i386_cpu_model(CPUX86State *env, char *cpu_model);
>          env = cpu_init();
> +        apply_i386_cpu_model(env, cpu_model);
>          if (i != 0)
>              env->hflags |= HF_HALTED_MASK;
>          if (smp_cpus > 1) {
> @@ -948,7 +950,7 @@
>      pc_init1(ram_size, vga_ram_size, boot_device,
>               ds, fd_filename, snapshot,
>               kernel_filename, kernel_cmdline,
> -             initrd_filename, 1);
> +             initrd_filename, 1, cpu_model);
>  }
> static void pc_init_isa(int ram_size, int vga_ram_size, int boot_device,
> @@ -962,7 +964,7 @@
>      pc_init1(ram_size, vga_ram_size, boot_device,
>               ds, fd_filename, snapshot,
>               kernel_filename, kernel_cmdline,
> -             initrd_filename, 0);
> +             initrd_filename, 0, cpu_model);
>  }
> QEMUMachine pc_machine = {
> Index: target-i386/cpu.h
> ===================================================================
> RCS file: /sources/qemu/qemu/target-i386/cpu.h,v
> retrieving revision 1.45
> diff -u -r1.45 cpu.h
> --- target-i386/cpu.h	11 Jul 2007 22:48:58 -0000	1.45
> +++ target-i386/cpu.h	10 Sep 2007 06:54:00 -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 -r1.48 helper2.c
> --- target-i386/helper2.c	31 Jul 2007 23:09:18 -0000	1.48
> +++ target-i386/helper2.c	10 Sep 2007 06:54:00 -0000
> @@ -45,40 +45,8 @@
>  #endif
>  #endif /* USE_CODE_COPY */
> -CPUX86State *cpu_x86_init(void)
> +static int set_guest_cpu_basic(CPUX86State *env)
>  {
> -    CPUX86State *env;
> -    static int inited;
> -
> -    env = qemu_mallocz(sizeof(CPUX86State));
> -    if (!env)
> -        return NULL;
> -    cpu_exec_init(env);
> -
> -    /* init various static tables */
> -    if (!inited) {
> -        inited = 1;
> -        optimize_flags_init();
> -    }
> -#ifdef USE_CODE_COPY
> -    /* testing code for code copy case */
> -    {
> -        struct modify_ldt_ldt_s ldt;
> -
> -        ldt.entry_number = 1;
> -        ldt.base_addr = (unsigned long)env;
> -        ldt.limit = (sizeof(CPUState) + 0xfff) >> 12;
> -        ldt.seg_32bit = 1;
> -        ldt.contents = MODIFY_LDT_CONTENTS_DATA;
> -        ldt.read_exec_only = 0;
> -        ldt.limit_in_pages = 1;
> -        ldt.seg_not_present = 0;
> -        ldt.useable = 1;
> -        modify_ldt(1, &ldt, sizeof(ldt)); /* write ldt entry */
> -
> -        asm volatile ("movl %0, %%fs" : : "r" ((1 << 3) | 7));
> -    }
> -#endif
>      {
>          int family, model, stepping;
>  #ifdef TARGET_X86_64
> @@ -114,19 +82,7 @@
>          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));
> -            }
> -        }
> +        env->cpuid_xlevel = 0x80000006;
>  #ifdef TARGET_X86_64
>          /* currently not enabled for std i386 because not fully tested  
> */
>          env->cpuid_ext2_features = (env->cpuid_features & 0x0183F3FF);
> @@ -139,6 +95,239 @@
>          env->cpuid_features |= CPUID_PSE36;
>  #endif
>      }
> +}
> +
> +static int host_cpuid (uint32_t function, uint32_t *ax,uint32_t *bx,  
> uint32_t *cx, uint32_t *dx)
> +{
> +    asm("cpuid"
> +        : "=a" (*ax),
> +          "=b" (*bx),
> +          "=c" (*cx),
> +          "=d" (*dx)
> +        : "a" (function));
> +}
> +
> +static int set_guest_cpu_hostlike(CPUX86State *env) {
> +    uint32_t ax, bx, cx, dx;
> +    host_cpuid(0, &ax, &bx, &cx, &dx);
> +    env->cpuid_level = 2; /* host real level is ax */
> +    env->cpuid_vendor1 = bx;
> +    env->cpuid_vendor2 = dx;
> +    env->cpuid_vendor3 = cx;
> +
> +    host_cpuid(1, &ax, &bx, &cx, &dx);
> +    env->cpuid_version = ax;
> +    env->cpuid_features = dx & ~CPUID_ACPI; /* acpi panics linux guests  
> */
> +    env->cpuid_ext_features = cx & ~CPUID_EXT_VMX; /* vmx is not  
> recursive */
> +
> +    /* cpuid(0x80000000, &ax, &bx, &cx, &dx); */
> +    env->cpuid_xlevel = 0x80000008;
> +
> +    host_cpuid(0x80000001, &ax, &bx, &cx, &dx);
> +    env->cpuid_ext2_features = dx;
> +
> +    return 0;
> +}
> +
> +/* 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);
> +}
> +
> +
> +void apply_i386_cpu_model(CPUX86State *env, char *cpu_model)
> +{
> +    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;
> +
> +    char *s, *featurestr;
> +
> +    if (!cpu_model)
> +        cpu_model = "host";
> +
> +    s = strdup(cpu_model);
> +
> +    strtok(s, ",");
> +    if (!strcmp(s, "basic"))
> +        set_guest_cpu_basic(env);
> +    else if (!strcmp(s, "host"))
> +        set_guest_cpu_hostlike(env);
> +    else {
> +        printf("Available CPUs:\n");
> +        printf("(basic|host)[,(+feature|-feature|feature=xyz)]...\n");
> +        exit(1);
> +    }
> +
> +    featurestr = strtok(NULL, ",");
> +    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;
> +
> +    {
> +        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));
> +        }
> +    }
> +}
> +
> +CPUX86State *cpu_x86_init(void)
> +{
> +    CPUX86State *env;
> +    static int inited;
> +
> +    env = qemu_mallocz(sizeof(CPUX86State));
> +    if (!env)
> +        return NULL;
> +    cpu_exec_init(env);
> +
> +    /* init various static tables */
> +    if (!inited) {
> +        inited = 1;
> +        optimize_flags_init();
> +    }
> +#ifdef USE_CODE_COPY
> +    /* testing code for code copy case */
> +    {
> +        struct modify_ldt_ldt_s ldt;
> +
> +        ldt.entry_number = 1;
> +        ldt.base_addr = (unsigned long)env;
> +        ldt.limit = (sizeof(CPUState) + 0xfff) >> 12;
> +        ldt.seg_32bit = 1;
> +        ldt.contents = MODIFY_LDT_CONTENTS_DATA;
> +        ldt.read_exec_only = 0;
> +        ldt.limit_in_pages = 1;
> +        ldt.seg_not_present = 0;
> +        ldt.useable = 1;
> +        modify_ldt(1, &ldt, sizeof(ldt)); /* write ldt entry */
> +
> +        asm volatile ("movl %0, %%fs" : : "r" ((1 << 3) | 7));
> +    }
> +#endif
>      cpu_reset(env);
>  #ifdef USE_KQEMU
>      kqemu_init(env);
>
>
>



-- 
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/