All of lore.kernel.org
 help / color / mirror / Atom feed
From: David Gibson <david@gibson.dropbear.id.au>
To: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
Cc: qemu-ppc@nongnu.org, qemu-devel@nongnu.org, clg@kaod.org,
	Benjamin Herrenschmidt <benh@kernel.crashing.org>
Subject: Re: [Qemu-devel] [PATCH v1 10/11] ppc/xics: Add "native" XICS subclass
Date: Mon, 27 Jun 2016 14:36:45 +1000	[thread overview]
Message-ID: <20160627043645.GI4242@voom.fritz.box> (raw)
In-Reply-To: <1466704050-15108-11-git-send-email-nikunj@linux.vnet.ibm.com>

[-- Attachment #1: Type: text/plain, Size: 14321 bytes --]

On Thu, Jun 23, 2016 at 11:17:29PM +0530, Nikunj A Dadhania wrote:
> From: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> 
> This provides MMIO based ICP access as found on POWER8
> 
> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
> Signed-off-by: Nikunj A Dadhania <nikunj@linux.vnet.ibm.com>
> ---
>  default-configs/ppc64-softmmu.mak |   3 +-
>  hw/intc/Makefile.objs             |   1 +
>  hw/intc/xics_native.c             | 295 ++++++++++++++++++++++++++++++++++++++
>  include/hw/ppc/xics.h             |  14 ++
>  4 files changed, 312 insertions(+), 1 deletion(-)
>  create mode 100644 hw/intc/xics_native.c
> 
> diff --git a/default-configs/ppc64-softmmu.mak b/default-configs/ppc64-softmmu.mak
> index c4be59f..315e30b 100644
> --- a/default-configs/ppc64-softmmu.mak
> +++ b/default-configs/ppc64-softmmu.mak
> @@ -48,8 +48,9 @@ CONFIG_PLATFORM_BUS=y
>  CONFIG_ETSEC=y
>  CONFIG_LIBDECNUMBER=y
>  # For pSeries
> -CONFIG_XICS=$(CONFIG_PSERIES)
> +CONFIG_XICS=$(or $(CONFIG_PSERIES),$(CONFIG_POWERNV))
>  CONFIG_XICS_SPAPR=$(CONFIG_PSERIES)
> +CONFIG_XICS_NATIVE=$(CONFIG_POWERNV)
>  CONFIG_XICS_KVM=$(and $(CONFIG_PSERIES),$(CONFIG_KVM))

I don't think you've introduced CONFIG_POWERNV in your patches so far,
making this a bit weird.

>  # For PReP
>  CONFIG_MC146818RTC=y
> diff --git a/hw/intc/Makefile.objs b/hw/intc/Makefile.objs
> index 530df2e..f8bbeda 100644
> --- a/hw/intc/Makefile.objs
> +++ b/hw/intc/Makefile.objs
> @@ -31,6 +31,7 @@ obj-$(CONFIG_RASPI) += bcm2835_ic.o bcm2836_control.o
>  obj-$(CONFIG_SH4) += sh_intc.o
>  obj-$(CONFIG_XICS) += xics.o
>  obj-$(CONFIG_XICS_SPAPR) += xics_spapr.o
> +obj-$(CONFIG_XICS_NATIVE) += xics_native.o
>  obj-$(CONFIG_XICS_KVM) += xics_kvm.o
>  obj-$(CONFIG_ALLWINNER_A10_PIC) += allwinner-a10-pic.o
>  obj-$(CONFIG_S390_FLIC) += s390_flic.o
> diff --git a/hw/intc/xics_native.c b/hw/intc/xics_native.c
> new file mode 100644
> index 0000000..26e45cc
> --- /dev/null
> +++ b/hw/intc/xics_native.c
> @@ -0,0 +1,295 @@
> +/*
> + * QEMU PowerPC hardware System Emulator
> + *
> + * Native version of ICS/ICP
> + *
> + * Copyright (c) 2010,2011 David Gibson, IBM Corporation.

Surely this should have yours or Ben's copyright in addition to mine.

> + * Permission is hereby granted, free of charge, to any person obtaining a copy
> + * of this software and associated documentation files (the "Software"), to deal
> + * in the Software without restriction, including without limitation the rights
> + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
> + * copies of the Software, and to permit persons to whom the Software is
> + * furnished to do so, subject to the following conditions:
> + *
> + * The above copyright notice and this permission notice shall be included in
> + * all copies or substantial portions of the Software.
> + *
> + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
> + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
> + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
> + * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
> + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
> + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
> + * THE SOFTWARE.
> + *
> + */
> +
> +#include "qemu/osdep.h"
> +#include "hw/hw.h"
> +#include "trace.h"
> +#include "qemu/timer.h"
> +#include "hw/ppc/xics.h"
> +#include "qapi/visitor.h"
> +#include "qapi/error.h"
> +
> +#include <libfdt.h>
> +
> +/* #define DEBUG_MM(fmt...)      printf(fmt) */
> +#define DEBUG_MM(fmt...)        do { } while (0)
> +
> +static void xics_native_initfn(Object *obj)
> +{
> +    XICSState *xics = XICS_NATIVE(obj);
> +
> +    QLIST_INIT(&xics->ics);
> +}
> +
> +static uint64_t icp_mm_read(void *opaque, hwaddr addr, unsigned width)
> +{
> +    XICSState *s = opaque;
> +    int32_t cpu_id, server;
> +    uint32_t val;
> +    ICPState *ss;
> +    bool byte0 = (width == 1 && (addr & 0x3) == 0);
> +
> +    cpu_id = (addr & (ICP_MM_SIZE - 1)) >> 12;
> +    server = get_cpu_index_by_dt_id(cpu_id);

Shouldn't each ICP instance register its own MMIO window?

> +    if (server < 0) {
> +        fprintf(stderr, "XICS: Bad ICP server %d\n", server);
> +        goto bad_access;
> +    }
> +    ss = &s->ss[server];
> +    switch (addr & 0xffc) {
> +    case 0: /* poll */
> +        val = icp_ipoll(ss, NULL);
> +        if (byte0) {
> +            val >>= 24;
> +        } else if (width != 4) {
> +            goto bad_access;
> +        }
> +        break;
> +    case 4: /* xirr */
> +        if (byte0) {
> +            val = icp_ipoll(ss, NULL) >> 24;
> +        } else if (width == 4) {
> +            val = icp_accept(ss);
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    case 12:
> +        if (byte0) {
> +            val = ss->mfrr;
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    case 16:
> +        if (width == 4) {
> +            val = ss->links[0];
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    case 20:
> +        if (width == 4) {
> +            val = ss->links[1];
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    case 24:
> +        if (width == 4) {
> +            val = ss->links[2];
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    default:
> +bad_access:
> +        fprintf(stderr, "XICS: Bad ICP access %llx/%d\n",
> +                (unsigned long long)addr, width);
> +        val = 0xffffffff;
> +    }
> +    DEBUG_MM("icp_mm_read(addr=%016llx,serv=0x%x/%d,off=%d,w=%d,val=0x%08x)\n",
> +             (unsigned long long)addr, cpu_id, server, (int)(addr & 0xffc),
> +             width, val);
> +
> +    return val;
> +}
> +
> +static void icp_mm_write(void *opaque, hwaddr addr, uint64_t val,
> +                        unsigned width)
> +{
> +    XICSState *s = opaque;
> +    int32_t cpu_id, server;
> +    ICPState *ss;
> +    bool byte0 = (width == 1 && (addr & 0x3) == 0);
> +
> +    cpu_id = (addr & (ICP_MM_SIZE - 1)) >> 12;
> +    server = get_cpu_index_by_dt_id(cpu_id);
> +    if (server < 0) {
> +        fprintf(stderr, "XICS: Bad ICP server %d\n", server);
> +        goto bad_access;
> +    }
> +    ss = &s->ss[server];
> +
> +    DEBUG_MM("icp_mm_write(addr=%016llx,serv=0x%x/%d,off=%d,w=%d,val=0x%08x)\n",
> +             (unsigned long long)addr, cpu_id, server,
> +             (int)(addr & 0xffc), width, (uint32_t)val);
> +
> +    switch (addr & 0xffc) {
> +    case 4: /* xirr */
> +        if (byte0) {
> +            icp_set_cppr(s, server, val);
> +        } else if (width == 4) {
> +            icp_eoi(s, server, val);
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    case 12:
> +        if (byte0) {
> +            icp_set_mfrr(s, server, val);
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    case 16:
> +        if (width == 4) {
> +            ss->links[0] = val;
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    case 20:
> +        if (width == 4) {
> +            ss->links[1] = val;
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    case 24:
> +        if (width == 4) {
> +            ss->links[2] = val;
> +        } else {
> +            goto bad_access;
> +        }
> +        break;
> +    default:
> + bad_access:
> +        val = 0xffffffff;
> +    }
> +}
> +
> +static const MemoryRegionOps icp_mm_ops = {
> +    .read = icp_mm_read,
> +    .write = icp_mm_write,
> +    .valid.min_access_size = 1,
> +    .valid.max_access_size = 4,
> +    .impl.min_access_size = 1,
> +    .impl.max_access_size = 4,
> +    .endianness = DEVICE_BIG_ENDIAN,
> +};
> +
> +#define _FDT(exp) \
> +    do { \
> +        int ret = (exp);                                           \
> +        if (ret < 0) {                                             \
> +            fprintf(stderr, "qemu: error creating device tree: %s: %s\n", \
> +                    #exp, fdt_strerror(ret));                      \
> +            exit(1);                                               \
> +        }                                                          \
> +    } while (0)

Ugh, I have got to find time to push by qdt cleanup stuff again.  Not
within scope for this patch obviously, though.

> +void xics_create_native_icp_node(XICSState *s, void *fdt,
> +                                 uint32_t base, uint32_t count)
> +{
> +    uint64_t addr;
> +    char *name;
> +    const char compat[] = "IBM,power8-icp\0IBM,ppc-xicp";
> +    uint32_t irange[2], i, rsize;
> +    uint64_t *reg;
> +
> +    addr = ICP_MM_BASE | (base << 12);
> +
> +    irange[0] = cpu_to_be32(base);
> +    irange[1] = cpu_to_be32(count);
> +
> +    rsize = sizeof(uint64_t) * 2 * count;
> +    reg = g_malloc(rsize);
> +    for (i = 0; i < count; i++) {
> +        reg[i * 2] = cpu_to_be64(addr | ((base + i) * 0x1000));
> +        reg[i * 2 + 1] = cpu_to_be64(0x1000);
> +    }
> +
> +    name = g_strdup_printf("interrupt-controller@%"PRIX64, addr);
> +
> +    /* interrupt controller */
> +    _FDT((fdt_begin_node(fdt, name)));
> +    g_free(name);
> +
> +    _FDT((fdt_property(fdt, "compatible", compat, sizeof(compat))));
> +    _FDT((fdt_property(fdt, "reg", reg, rsize)));
> +    _FDT((fdt_property_string(fdt, "device_type",
> +                              "PowerPC-External-Interrupt-Presentation")));
> +    _FDT((fdt_property(fdt, "interrupt-controller", NULL, 0)));
> +    _FDT((fdt_property(fdt, "ibm,interrupt-server-ranges",
> +                       irange, sizeof(irange))));
> +    _FDT((fdt_property_cell(fdt, "#interrupt-cells", 1)));
> +    _FDT((fdt_property_cell(fdt, "#address-cells", 0)));
> +    _FDT((fdt_end_node(fdt)));
> +}
> +
> +static void xics_native_realize(DeviceState *dev, Error **errp)
> +{
> +    XICSState *s = XICS_NATIVE(dev);
> +    SysBusDevice *sbd = SYS_BUS_DEVICE(dev);
> +    Error *error = NULL;
> +    int i;
> +
> +    if (!s->nr_servers) {
> +        error_setg(errp, "Number of servers needs to be greater 0");
> +        return;
> +    }
> +
> +    /* Register MMIO regions */
> +    memory_region_init_io(&s->icp_mmio, OBJECT(s), &icp_mm_ops, s, "icp",
> +                          ICP_MM_SIZE);
> +    sysbus_init_mmio(sbd, &s->icp_mmio);
> +    sysbus_mmio_map(sbd, 0, ICP_MM_BASE);
> +
> +    for (i = 0; i < s->nr_servers; i++) {
> +        object_property_set_bool(OBJECT(&s->ss[i]), true, "realized", &error);
> +        if (error) {
> +            error_propagate(errp, error);
> +            return;
> +        }
> +    }
> +}
> +
> +static void xics_native_class_init(ObjectClass *oc, void *data)
> +{
> +    DeviceClass *dc = DEVICE_CLASS(oc);
> +    XICSStateClass *xsc = XICS_NATIVE_CLASS(oc);
> +
> +    dc->realize = xics_native_realize;
> +    xsc->set_nr_servers = xics_set_nr_servers;
> +}
> +
> +static const TypeInfo xics_native_info = {
> +    .name          = TYPE_XICS_NATIVE,
> +    .parent        = TYPE_XICS_COMMON,
> +    .instance_size = sizeof(XICSState),
> +    .class_size = sizeof(XICSStateClass),
> +    .class_init    = xics_native_class_init,
> +    .instance_init = xics_native_initfn,
> +};
> +
> +static void xics_native_register_types(void)
> +{
> +    type_register_static(&xics_native_info);
> +}
> +type_init(xics_native_register_types)
> diff --git a/include/hw/ppc/xics.h b/include/hw/ppc/xics.h
> index 6fb1cb4..0e47ecb 100644
> --- a/include/hw/ppc/xics.h
> +++ b/include/hw/ppc/xics.h
> @@ -42,6 +42,9 @@
>  #define TYPE_XICS_SPAPR_KVM "xics-spapr-kvm"
>  #define KVM_XICS(obj) OBJECT_CHECK(KVMXICSState, (obj), TYPE_XICS_SPAPR_KVM)
>  
> +#define TYPE_XICS_NATIVE "xics-native"
> +#define XICS_NATIVE(obj) OBJECT_CHECK(XICSState, (obj), TYPE_XICS_NATIVE)
> +
>  #define XICS_COMMON_CLASS(klass) \
>       OBJECT_CLASS_CHECK(XICSStateClass, (klass), TYPE_XICS_COMMON)
>  #define XICS_SPAPR_CLASS(klass) \
> @@ -50,6 +53,8 @@
>       OBJECT_GET_CLASS(XICSStateClass, (obj), TYPE_XICS_COMMON)
>  #define XICS_SPAPR_GET_CLASS(obj) \
>       OBJECT_GET_CLASS(XICSStateClass, (obj), TYPE_XICS_SPAPR)
> +#define XICS_NATIVE_CLASS(klass) \
> +     OBJECT_CLASS_CHECK(XICSStateClass, (klass), TYPE_XICS_NATIVE)
>  
>  #define XICS_IPI        0x2
>  #define XICS_BUID       0x1
> @@ -84,6 +89,7 @@ struct XICSState {
>      uint32_t nr_irqs;
>      ICPState *ss;
>      QLIST_HEAD(, ICSState) ics;
> +    MemoryRegion icp_mmio;

This is only used by XICSNative, so it should be in a structure just
for the subclass, not in the common struct.

>  };
>  
>  #define TYPE_ICP "icp"
> @@ -115,8 +121,13 @@ struct ICPState {
>      uint8_t mfrr;
>      qemu_irq output;
>      bool cap_irq_xics_enabled;
> +    uint32_t links[3];

Likewise here.

>  };
>  
> +/* This should be an XSCOM BAR ... the size is arbitrary as well */
> +#define ICP_MM_BASE     0x0003FFFF80000000
> +#define ICP_MM_SIZE     0x0000000010000000
> +
>  #define TYPE_ICS_BASE "ics-base"
>  #define ICS_BASE(obj) OBJECT_CHECK(ICSState, (obj), TYPE_ICS_SIMPLE)
>  
> @@ -187,6 +198,9 @@ void xics_spapr_free(XICSState *icp, int irq, int num);
>  void xics_cpu_setup(XICSState *icp, PowerPCCPU *cpu);
>  void xics_cpu_destroy(XICSState *icp, PowerPCCPU *cpu);
>  
> +void xics_create_native_icp_node(XICSState *s, void *fdt,
> +                                 uint32_t base, uint32_t count);
> +
>  /* Internal XICS interfaces */
>  int get_cpu_index_by_dt_id(int cpu_dt_id);
>  

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

  reply	other threads:[~2016-06-27  4:47 UTC|newest]

Thread overview: 37+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-06-23 17:47 [Qemu-devel] [PATCH v1 00/11] sPAPR xics rework/cleanup Nikunj A Dadhania
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 01/11] ppc/xics: Rename existing xics to xics_spapr Nikunj A Dadhania
2016-06-24  5:17   ` David Gibson
2016-06-24  5:53     ` Nikunj A Dadhania
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 02/11] ppc/xics: Move SPAPR specific code to a separate file Nikunj A Dadhania
2016-06-24  5:19   ` David Gibson
2016-06-24  5:57     ` Nikunj A Dadhania
2016-06-24  6:12       ` David Gibson
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 03/11] ppc/xics: Implement H_IPOLL using an accessor Nikunj A Dadhania
2016-06-24  5:21   ` David Gibson
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 04/11] ppc/xics: Remove unused xics_set_irq_type() Nikunj A Dadhania
2016-06-24  5:45   ` David Gibson
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 05/11] ppc/xics: Replace "icp" with "xics" in most places Nikunj A Dadhania
2016-06-27  3:43   ` David Gibson
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 06/11] ppc/xics: Make the ICSState a list Nikunj A Dadhania
2016-06-27  4:20   ` David Gibson
2016-06-27  4:56     ` Nikunj A Dadhania
2016-06-27  5:30       ` David Gibson
2016-06-27  5:32         ` Nikunj A Dadhania
2016-06-27  4:59     ` Nikunj A Dadhania
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 07/11] ppc/xics: An ICS with offset 0 is assumed to be uninitialized Nikunj A Dadhania
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 08/11] ppc/xics: Use a helper to add a new ICS Nikunj A Dadhania
2016-06-27  4:21   ` David Gibson
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 09/11] ppc/xics: Split ICS into ics-base and ics class Nikunj A Dadhania
2016-06-27  4:26   ` David Gibson
2016-06-27  5:18     ` Nikunj A Dadhania
2016-06-27 10:11       ` Nikunj A Dadhania
2016-06-28  3:00         ` David Gibson
2016-06-28  5:06           ` Nikunj A Dadhania
2016-06-28  6:25             ` David Gibson
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 10/11] ppc/xics: Add "native" XICS subclass Nikunj A Dadhania
2016-06-27  4:36   ` David Gibson [this message]
2016-06-27  9:53     ` Nikunj A Dadhania
2016-06-28  2:58       ` David Gibson
2016-06-23 17:47 ` [Qemu-devel] [PATCH v1 11/11] ppc/xics: Add xics to the monitor "info pic" command Nikunj A Dadhania
2016-06-27  4:48   ` David Gibson
2016-06-27  5:56     ` Benjamin Herrenschmidt

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=20160627043645.GI4242@voom.fritz.box \
    --to=david@gibson.dropbear.id.au \
    --cc=benh@kernel.crashing.org \
    --cc=clg@kaod.org \
    --cc=nikunj@linux.vnet.ibm.com \
    --cc=qemu-devel@nongnu.org \
    --cc=qemu-ppc@nongnu.org \
    /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: link
Be 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.