Re: [Xen-devel] [PATCH v2 1/8] x86: detect and initialize Cache QoS Monitoring feature

["Xu, Dongxiao" <dongxiao.xu@xxxxxxxxx>]

> -----Original Message-----
> From: Xu, Dongxiao
> Sent: Monday, November 25, 2013 11:07 AM
> To: Andrew Cooper
> Cc: xen-devel@xxxxxxxxxxxxx
> Subject: RE: [Xen-devel] [PATCH v2 1/8] x86: detect and initialize Cache QoS
> Monitoring feature
> 
> > -----Original Message-----
> > From: Andrew Cooper [mailto:andrew.cooper3@xxxxxxxxxx]
> > Sent: Thursday, November 21, 2013 8:14 PM
> > To: Xu, Dongxiao
> > Cc: xen-devel@xxxxxxxxxxxxx
> > Subject: Re: [Xen-devel] [PATCH v2 1/8] x86: detect and initialize Cache QoS
> > Monitoring feature
> >
> > On 21/11/13 07:20, dongxiao.xu@xxxxxxxxx wrote:
> > > From: Dongxiao Xu <dongxiao.xu@xxxxxxxxx>
> > >
> > > Detect platform QoS feature status and enumerate the resource types,
> > > one of which is to monitor the L3 cache occupancy.
> > >
> > > Also introduce a Xen grub command line parameter to control the
> > > QoS feature status globally.
> > >
> > > Signed-off-by: Jiongxi Li <jiongxi.li@xxxxxxxxx>
> > > Signed-off-by: Dongxiao Xu <dongxiao.xu@xxxxxxxxx>
> > > ---
> > >  xen/arch/x86/Makefile            |    1 +
> > >  xen/arch/x86/cpu/intel.c         |    6 +++
> > >  xen/arch/x86/pqos.c              |   89
> > ++++++++++++++++++++++++++++++++++++++
> > >  xen/arch/x86/setup.c             |    3 ++
> > >  xen/include/asm-x86/cpufeature.h |    1 +
> > >  xen/include/asm-x86/pqos.h       |   37 ++++++++++++++++
> > >  6 files changed, 137 insertions(+)
> > >  create mode 100644 xen/arch/x86/pqos.c
> > >  create mode 100644 xen/include/asm-x86/pqos.h
> > >
> > > diff --git a/xen/arch/x86/Makefile b/xen/arch/x86/Makefile
> > > index d502bdf..54962e0 100644
> > > --- a/xen/arch/x86/Makefile
> > > +++ b/xen/arch/x86/Makefile
> > > @@ -58,6 +58,7 @@ obj-y += crash.o
> > >  obj-y += tboot.o
> > >  obj-y += hpet.o
> > >  obj-y += xstate.o
> > > +obj-y += pqos.o
> > >
> > >  obj-$(crash_debug) += gdbstub.o
> > >
> > > diff --git a/xen/arch/x86/cpu/intel.c b/xen/arch/x86/cpu/intel.c
> > > index 27fe762..f0d83ea 100644
> > > --- a/xen/arch/x86/cpu/intel.c
> > > +++ b/xen/arch/x86/cpu/intel.c
> > > @@ -230,6 +230,12 @@ static void __devinit init_intel(struct cpuinfo_x86 
> > > *c)
> > >        ( c->cpuid_level >= 0x00000006 ) &&
> > >        ( cpuid_eax(0x00000006) & (1u<<2) ) )
> > >           set_bit(X86_FEATURE_ARAT, c->x86_capability);
> > > +
> > > + /* Check platform QoS monitoring capability */
> > > + if ((c->cpuid_level >= 0x00000007) &&
> > > +     (cpuid_ebx(0x00000007) & (1u<<12)))
> > > +         set_bit(X86_FEATURE_QOSM, c->x86_capability);
> > > +
> > >  }
> > >
> > >  static struct cpu_dev intel_cpu_dev __cpuinitdata = {
> > > diff --git a/xen/arch/x86/pqos.c b/xen/arch/x86/pqos.c
> > > new file mode 100644
> > > index 0000000..e6ab416
> > > --- /dev/null
> > > +++ b/xen/arch/x86/pqos.c
> > > @@ -0,0 +1,89 @@
> > > +/*
> > > + * pqos.c: Platform QoS related service for guest.
> > > + *
> > > + * Copyright (c) 2013, Intel Corporation
> > > + * Author: Jiongxi Li  <jiongxi.li@xxxxxxxxx>
> > > + * Author: Dongxiao Xu <dongxiao.xu@xxxxxxxxx>
> > > + *
> > > + * This program is free software; you can redistribute it and/or modify 
> > > it
> > > + * under the terms and conditions of the GNU General Public License,
> > > + * version 2, as published by the Free Software Foundation.
> > > + *
> > > + * This program is distributed in the hope it will be useful, but WITHOUT
> > > + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
> > or
> > > + * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
> > License for
> > > + * more details.
> > > + *
> > > + * You should have received a copy of the GNU General Public License 
> > > along
> > with
> > > + * this program; if not, write to the Free Software Foundation, Inc., 59
> Temple
> > > + * Place - Suite 330, Boston, MA 02111-1307 USA.
> > > + */
> > > +#include <asm/processor.h>
> > > +#include <xen/init.h>
> > > +#include <asm/pqos.h>
> > > +
> > > +static bool_t pqos_enabled = 1;
> >
> > I cant see anywhere in the series where this is used outside of an init
> > function, in which case it should be __initdata.
> 
> OK.
> 
> >
> > > +boolean_param("pqos", pqos_enabled);
> > > +
> > > +unsigned int cqm_res_count = 0;
> > > +unsigned int cqm_upscaling_factor = 0;
> > > +bool_t cqm_enabled = 0;
> > > +struct cqm_res_struct *cqm_res_array = NULL;
> > > +
> > > +static void __init init_cqm(void)
> > > +{
> > > +    unsigned int eax, edx;
> > > +    unsigned int max_cqm_rmid;
> > > +
> > > +    cpuid_count(0xf, 1, &eax, &cqm_upscaling_factor, &max_cqm_rmid,
> > &edx);
> > > +    if ( !(edx & QOS_MONITOR_EVTID_L3) )
> > > +        return;
> > > +
> > > +    cqm_res_count = max_cqm_rmid + 1;
> > > +
> >
> > Range check on cqm_res_count ?  If max_cqm_rmid ends up as -1 from the
> > cpuid, we will allocate a 0 length array and crash later when reserving
> > RMID 0
> 
> That's a good point.
> I will add a range check here.

According to the SDM, the biggest RMID value should be 0xffffffff (if 
possible), so the biggest cqm_res_count is 0x100000000 (if possible).

So what about define cqm_res_count as "unsigned long"?

Thanks,
Dongxiao

> 
> >
> > > +    cqm_res_array = xzalloc_array(struct cqm_res_struct, cqm_res_count);
> > > +    if ( !cqm_res_array )
> > > +    {
> > > +        cqm_res_count = 0;
> > > +        return;
> > > +    }
> > > +
> > > +    cqm_enabled = 1;
> > > +
> > > +    /* Reserve RMID 0 for all domains not being monitored */
> > > +    cqm_res_array[0].inuse = 1;
> > > +
> > > +    printk(XENLOG_INFO "Cache QoS Monitoring Enabled.\n");
> > > +}
> > > +
> > > +static void __init init_qos_monitor(void)
> > > +{
> > > +    unsigned int qm_features;
> > > +    unsigned int eax, ebx, ecx;
> > > +
> > > +    if ( !(boot_cpu_has(X86_FEATURE_QOSM)) )
> > > +        return;
> > > +
> > > +    cpuid_count(0xf, 0, &eax, &ebx, &ecx, &qm_features);
> > > +
> > > +    if ( qm_features & QOS_MONITOR_TYPE_L3 )
> > > +        init_cqm();
> > > +}
> > > +
> > > +void __init init_platform_qos(void)
> > > +{
> > > +    if ( !pqos_enabled )
> > > +        return;
> > > +
> > > +    init_qos_monitor();
> > > +}
> > > +
> > > +/*
> > > + * Local variables:
> > > + * mode: C
> > > + * c-file-style: "BSD"
> > > + * c-basic-offset: 4
> > > + * tab-width: 4
> > > + * indent-tabs-mode: nil
> > > + * End:
> > > + */
> > > diff --git a/xen/arch/x86/setup.c b/xen/arch/x86/setup.c
> > > index 5bf4ee0..95418e4 100644
> > > --- a/xen/arch/x86/setup.c
> > > +++ b/xen/arch/x86/setup.c
> > > @@ -48,6 +48,7 @@
> > >  #include <asm/setup.h>
> > >  #include <xen/cpu.h>
> > >  #include <asm/nmi.h>
> > > +#include <asm/pqos.h>
> > >
> > >  /* opt_nosmp: If true, secondary processors are ignored. */
> > >  static bool_t __initdata opt_nosmp;
> > > @@ -1402,6 +1403,8 @@ void __init __start_xen(unsigned long mbi_p)
> > >
> > >      domain_unpause_by_systemcontroller(dom0);
> > >
> > > +    init_platform_qos();
> > > +
> >
> > Does this want to be slightly earlier during startup? Does it make sense
> > to have dom0 monitored?  Is it possible to monitor Xen's L3 usage as well?
> 
> In this design, we monitored per domain basis, including all the L3 cache 
> usage
> within both guest and hypervisor mode.
> Dom0 can also be monitored by "xl pqos-attach cqm 0".
> 
> However for specific Xen's L3 usage, it is not covered in this design, since 
> we can
> only write one RMID into the hardware at a time.
> 
> >
> > >      reset_stack_and_jump(init_done);
> > >  }
> > >
> > > diff --git a/xen/include/asm-x86/cpufeature.h
> > b/xen/include/asm-x86/cpufeature.h
> > > index 1cfaf94..ca59668 100644
> > > --- a/xen/include/asm-x86/cpufeature.h
> > > +++ b/xen/include/asm-x86/cpufeature.h
> > > @@ -147,6 +147,7 @@
> > >  #define X86_FEATURE_ERMS (7*32+ 9) /* Enhanced REP MOVSB/STOSB */
> > >  #define X86_FEATURE_INVPCID      (7*32+10) /* Invalidate Process Context
> > ID */
> > >  #define X86_FEATURE_RTM  (7*32+11) /* Restricted Transactional
> > Memory */
> > > +#define X86_FEATURE_QOSM (7*32+12) /* Platform QoS monitoring
> > capability */
> > >  #define X86_FEATURE_NO_FPU_SEL   (7*32+13) /* FPU CS/DS stored as
> > zero */
> > >  #define X86_FEATURE_SMAP (7*32+20) /* Supervisor Mode Access
> > Prevention */
> > >
> > > diff --git a/xen/include/asm-x86/pqos.h b/xen/include/asm-x86/pqos.h
> > > new file mode 100644
> > > index 0000000..934d68a
> > > --- /dev/null
> > > +++ b/xen/include/asm-x86/pqos.h
> >
> > It might make sense to split this file.  The QOS_* defines could be
> > argued as asm-x86, (perhaps asm-x86/qpos-defines.h)
> >
> > However, the cqm_res_struct and function declaration are very certainly
> > not asm code, and would probably better live in arch/x86/qpos.h
> >
> > > @@ -0,0 +1,37 @@
> > > +/*
> > > + * pqos.h: Platform QoS related service for guest.
> > > + *
> > > + * Copyright (c) 2013, Intel Corporation
> > > + * Author: Jiongxi Li  <jiongxi.li@xxxxxxxxx>
> > > + * Author: Dongxiao Xu <dongxiao.xu@xxxxxxxxx>
> > > + *
> > > + * This program is free software; you can redistribute it and/or modify 
> > > it
> > > + * under the terms and conditions of the GNU General Public License,
> > > + * version 2, as published by the Free Software Foundation.
> > > + *
> > > + * This program is distributed in the hope it will be useful, but WITHOUT
> > > + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
> > or
> > > + * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public
> > License for
> > > + * more details.
> > > + *
> > > + * You should have received a copy of the GNU General Public License 
> > > along
> > with
> > > + * this program; if not, write to the Free Software Foundation, Inc., 59
> Temple
> > > + * Place - Suite 330, Boston, MA 02111-1307 USA.
> > > + */
> > > +#ifndef ASM_PQOS_H
> > > +#define ASM_PQOS_H
> > > +
> > > +/* QoS Resource Type Enumeration */
> > > +#define QOS_MONITOR_TYPE_L3            0x2
> > > +
> > > +/* QoS Monitoring Event ID */
> > > +#define QOS_MONITOR_EVTID_L3           0x1
> > > +
> > > +struct cqm_res_struct {
> >
> > This name is redundant with the extra "_struct" at the end. How about
> > cqm_res or cqm_resource ?
> 
> OK.
> 
> >
> > > +    domid_t  domain_id;
> >
> > You will need to include one of the types headers.
> 
> Sorry maybe I didn't quite understand your point here... But, domid_t's 
> definition
> header file will be included in the arch/x86/pqos.c file before pqos.h is 
> included.
> 
> >
> > > +    bool_t   inuse;
> > > +};
> > > +
> > > +void init_platform_qos(void);
> > > +
> > > +#endif


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel