qemu-ppc
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [PATCH v6] ppc: Enable 2nd DAWR support on p10


From: Daniel Henrique Barboza
Subject: Re: [PATCH v6] ppc: Enable 2nd DAWR support on p10
Date: Fri, 7 Jul 2023 11:49:50 -0300
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101 Thunderbird/102.12.0



On 7/7/23 11:24, Cédric Le Goater wrote:
On 7/7/23 13:59, Greg Kurz wrote:
Hi Daniel and Shiva !

On Fri, 7 Jul 2023 08:09:47 -0300
Daniel Henrique Barboza <danielhb413@gmail.com> wrote:

This one was a buzzer shot.


Indeed ! :-) I would have appreciated some more time to re-assess
my R-b tag on this 2 year old bug though ;-)

We should drop that patch IMO and ask for a resend with more tests
but that's a lot of work to build a PR :/


I don't mind the extra work.

Patch dropped. Feel free to re-send it for 8.2 with more tests as Cedric
suggested.



Thanks,


Daniel



C.



My concerns were that the DAWR1 spapr cap was still not enabled by
default but I guess it is because POWER9 is still the default cpu
type. Related, the apply function should probably spit a warning
with TCG instead of failing, like already done for some other
TCG limitations (e.g. cap_safe_bounds_check_apply()). This will
be needed for `make test` to succeed when DAWR1 is eventually
enabled by default. Not needed right now.

My R-b still stands then ! :-)

Cheers,

--
Greg


Queued in gitlab.com/danielhb/qemu/tree/ppc-next. Thanks,


Daniel


On 7/7/23 05:47, Shivaprasad G Bhat wrote:
From: Ravi Bangoria <ravi.bangoria@linux.ibm.com>

As per the PAPR, bit 0 of byte 64 in pa-features property
indicates availability of 2nd DAWR registers. i.e. If this bit is set, 2nd
DAWR is present, otherwise not. Use KVM_CAP_PPC_DAWR1 capability to find
whether kvm supports 2nd DAWR or not. If it's supported, allow user to set
the pa-feature bit in guest DT using cap-dawr1 machine capability. Though,
watchpoint on powerpc TCG guest is not supported and thus 2nd DAWR is not
enabled for TCG mode.

Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Signed-off-by: Shivaprasad G Bhat <sbhat@linux.ibm.com>
---
Changelog:
v5: 
https://lore.kernel.org/all/20210412114433.129702-1-ravi.bangoria@linux.ibm.com/
v5->v6:
    - The other patches in the original series already merged.
    - Rebased to the top of the tree. So, the gen_spr_book3s_310_dbg() is 
renamed
      to register_book3s_310_dbg_sprs() and moved to cpu_init.c accordingly.
    - No functional changes.

v4: 
https://lore.kernel.org/r/20210406053833.282907-1-ravi.bangoria@linux.ibm.com
v3->v4:
    - Make error message more proper.

v3: https://lore.kernel.org/r/20210330095350.36309-1-ravi.bangoria@linux.ibm.com
v3->v4:
    - spapr_dt_pa_features(): POWER10 processor is compatible with 3.0
      (PCR_COMPAT_3_00). No need to ppc_check_compat(3_10) for now as
      ppc_check_compati(3_00) will also be true. ppc_check_compat(3_10)
      can be added while introducing pa_features_310 in future.
    - Use error_append_hint() for hints. Also add ERRP_GUARD().
    - Add kvmppc_set_cap_dawr1() stub function for CONFIG_KVM=n.

v2: 
https://lore.kernel.org/r/20210329041906.213991-1-ravi.bangoria@linux.ibm.com
v2->v3:
    - Don't introduce pa_features_310[], instead, reuse pa_features_300[]
      for 3.1 guests, as there is no difference between initial values of
      them atm.
    - Call gen_spr_book3s_310_dbg() from init_proc_POWER10() instead of
      init_proc_POWER8(). Also, Don't call gen_spr_book3s_207_dbg() from
      gen_spr_book3s_310_dbg() as init_proc_POWER10() already calls it.

v1: 
https://lore.kernel.org/r/20200723104220.314671-1-ravi.bangoria@linux.ibm.com
v1->v2:
    - Introduce machine capability cap-dawr1 to enable/disable
      the feature. By default, 2nd DAWR is OFF for guests even
      when host kvm supports it. User has to manually enable it
      with -machine cap-dawr1=on if he wishes to use it.
    - Split the header file changes into separate patch. (Sync
      headers from v5.12-rc3)

[1] https://git.kernel.org/torvalds/c/bd1de1a0e6eff

   hw/ppc/spapr.c         |    7 ++++++-
   hw/ppc/spapr_caps.c    |   32 ++++++++++++++++++++++++++++++++
   include/hw/ppc/spapr.h |    6 +++++-
   target/ppc/cpu.h       |    2 ++
   target/ppc/cpu_init.c  |   15 +++++++++++++++
   target/ppc/kvm.c       |   12 ++++++++++++
   target/ppc/kvm_ppc.h   |   12 ++++++++++++
   7 files changed, 84 insertions(+), 2 deletions(-)

diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
index 54dbfd7fe9..1e54e0c719 100644
--- a/hw/ppc/spapr.c
+++ b/hw/ppc/spapr.c
@@ -241,7 +241,7 @@ static void spapr_dt_pa_features(SpaprMachineState *spapr,
           0x80, 0x00, 0x80, 0x00, 0x80, 0x00, /* 48 - 53 */
           /* 54: DecFP, 56: DecI, 58: SHA */
           0x80, 0x00, 0x80, 0x00, 0x80, 0x00, /* 54 - 59 */
-        /* 60: NM atomic, 62: RNG */
+        /* 60: NM atomic, 62: RNG, 64: DAWR1 (ISA 3.1) */
           0x80, 0x00, 0x80, 0x00, 0x00, 0x00, /* 60 - 65 */
       };
       uint8_t *pa_features = NULL;
@@ -282,6 +282,9 @@ static void spapr_dt_pa_features(SpaprMachineState *spapr,
            * in pa-features. So hide it from them. */
           pa_features[40 + 2] &= ~0x80; /* Radix MMU */
       }
+    if (spapr_get_cap(spapr, SPAPR_CAP_DAWR1)) {
+        pa_features[66] |= 0x80;
+    }

       _FDT((fdt_setprop(fdt, offset, "ibm,pa-features", pa_features, 
pa_size)));
   }
@@ -2084,6 +2087,7 @@ static const VMStateDescription vmstate_spapr = {
           &vmstate_spapr_cap_fwnmi,
           &vmstate_spapr_fwnmi,
           &vmstate_spapr_cap_rpt_invalidate,
+        &vmstate_spapr_cap_dawr1,
           NULL
       }
   };
@@ -4683,6 +4687,7 @@ static void spapr_machine_class_init(ObjectClass *oc, 
void *data)
       smc->default_caps.caps[SPAPR_CAP_CCF_ASSIST] = SPAPR_CAP_ON;
       smc->default_caps.caps[SPAPR_CAP_FWNMI] = SPAPR_CAP_ON;
       smc->default_caps.caps[SPAPR_CAP_RPT_INVALIDATE] = SPAPR_CAP_OFF;
+    smc->default_caps.caps[SPAPR_CAP_DAWR1] = SPAPR_CAP_OFF;

       /*
        * This cap specifies whether the AIL 3 mode for
diff --git a/hw/ppc/spapr_caps.c b/hw/ppc/spapr_caps.c
index 5a0755d34f..2f2cf4a250 100644
--- a/hw/ppc/spapr_caps.c
+++ b/hw/ppc/spapr_caps.c
@@ -655,6 +655,28 @@ static void cap_ail_mode_3_apply(SpaprMachineState *spapr,
       }
   }

+static void cap_dawr1_apply(SpaprMachineState *spapr, uint8_t val,
+                               Error **errp)
+{
+    ERRP_GUARD();
+    if (!val) {
+        return; /* Disable by default */
+    }
+
+    if (tcg_enabled()) {
+        error_setg(errp, "DAWR1 not supported in TCG.");
+        error_append_hint(errp, "Try appending -machine cap-dawr1=off\n");
+    } else if (kvm_enabled()) {
+        if (!kvmppc_has_cap_dawr1()) {
+            error_setg(errp, "DAWR1 not supported by KVM.");
+            error_append_hint(errp, "Try appending -machine cap-dawr1=off\n");
+        } else if (kvmppc_set_cap_dawr1(val) < 0) {
+            error_setg(errp, "Error enabling cap-dawr1 with KVM.");
+            error_append_hint(errp, "Try appending -machine cap-dawr1=off\n");
+        }
+    }
+}
+
   SpaprCapabilityInfo capability_table[SPAPR_CAP_NUM] = {
       [SPAPR_CAP_HTM] = {
           .name = "htm",
@@ -781,6 +803,15 @@ SpaprCapabilityInfo capability_table[SPAPR_CAP_NUM] = {
           .type = "bool",
           .apply = cap_ail_mode_3_apply,
       },
+    [SPAPR_CAP_DAWR1] = {
+        .name = "dawr1",
+        .description = "Allow 2nd Data Address Watchpoint Register (DAWR1)",
+        .index = SPAPR_CAP_DAWR1,
+        .get = spapr_cap_get_bool,
+        .set = spapr_cap_set_bool,
+        .type = "bool",
+        .apply = cap_dawr1_apply,
+    },
   };

   static SpaprCapabilities default_caps_with_cpu(SpaprMachineState *spapr,
@@ -923,6 +954,7 @@ SPAPR_CAP_MIG_STATE(large_decr, 
SPAPR_CAP_LARGE_DECREMENTER);
   SPAPR_CAP_MIG_STATE(ccf_assist, SPAPR_CAP_CCF_ASSIST);
   SPAPR_CAP_MIG_STATE(fwnmi, SPAPR_CAP_FWNMI);
   SPAPR_CAP_MIG_STATE(rpt_invalidate, SPAPR_CAP_RPT_INVALIDATE);
+SPAPR_CAP_MIG_STATE(dawr1, SPAPR_CAP_DAWR1);

   void spapr_caps_init(SpaprMachineState *spapr)
   {
diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h
index 538b2dfb89..47fffb921a 100644
--- a/include/hw/ppc/spapr.h
+++ b/include/hw/ppc/spapr.h
@@ -80,8 +80,10 @@ typedef enum {
   #define SPAPR_CAP_RPT_INVALIDATE        0x0B
   /* Support for AIL modes */
   #define SPAPR_CAP_AIL_MODE_3            0x0C
+/* DAWR1 */
+#define SPAPR_CAP_DAWR1                 0x0D
   /* Num Caps */
-#define SPAPR_CAP_NUM                   (SPAPR_CAP_AIL_MODE_3 + 1)
+#define SPAPR_CAP_NUM                   (SPAPR_CAP_DAWR1 + 1)

   /*
    * Capability Values
@@ -407,6 +409,7 @@ struct SpaprMachineState {
   #define H_SET_MODE_RESOURCE_SET_DAWR0           2
   #define H_SET_MODE_RESOURCE_ADDR_TRANS_MODE     3
   #define H_SET_MODE_RESOURCE_LE                  4
+#define H_SET_MODE_RESOURCE_SET_DAWR1           5

   /* Flags for H_SET_MODE_RESOURCE_LE */
   #define H_SET_MODE_ENDIAN_BIG    0
@@ -990,6 +993,7 @@ extern const VMStateDescription 
vmstate_spapr_cap_ccf_assist;
   extern const VMStateDescription vmstate_spapr_cap_fwnmi;
   extern const VMStateDescription vmstate_spapr_cap_rpt_invalidate;
   extern const VMStateDescription vmstate_spapr_wdt;
+extern const VMStateDescription vmstate_spapr_cap_dawr1;

   static inline uint8_t spapr_get_cap(SpaprMachineState *spapr, int cap)
   {
diff --git a/target/ppc/cpu.h b/target/ppc/cpu.h
index af12c93ebc..64855935f7 100644
--- a/target/ppc/cpu.h
+++ b/target/ppc/cpu.h
@@ -1588,9 +1588,11 @@ void ppc_compat_add_property(Object *obj, const char 
*name,
   #define SPR_PSPB              (0x09F)
   #define SPR_DPDES             (0x0B0)
   #define SPR_DAWR0             (0x0B4)
+#define SPR_DAWR1             (0x0B5)
   #define SPR_RPR               (0x0BA)
   #define SPR_CIABR             (0x0BB)
   #define SPR_DAWRX0            (0x0BC)
+#define SPR_DAWRX1            (0x0BD)
   #define SPR_HFSCR             (0x0BE)
   #define SPR_VRSAVE            (0x100)
   #define SPR_USPRG0            (0x100)
diff --git a/target/ppc/cpu_init.c b/target/ppc/cpu_init.c
index aeff71d063..c688e52928 100644
--- a/target/ppc/cpu_init.c
+++ b/target/ppc/cpu_init.c
@@ -5131,6 +5131,20 @@ static void register_book3s_207_dbg_sprs(CPUPPCState 
*env)
                           KVM_REG_PPC_CIABR, 0x00000000);
   }

+static void register_book3s_310_dbg_sprs(CPUPPCState *env)
+{
+    spr_register_kvm_hv(env, SPR_DAWR1, "DAWR1",
+                        SPR_NOACCESS, SPR_NOACCESS,
+                        SPR_NOACCESS, SPR_NOACCESS,
+                        &spr_read_generic, &spr_write_generic,
+                        KVM_REG_PPC_DAWR1, 0x00000000);
+    spr_register_kvm_hv(env, SPR_DAWRX1, "DAWRX1",
+                        SPR_NOACCESS, SPR_NOACCESS,
+                        SPR_NOACCESS, SPR_NOACCESS,
+                        &spr_read_generic, &spr_write_generic32,
+                        KVM_REG_PPC_DAWRX1, 0x00000000);
+}
+
   static void register_970_dbg_sprs(CPUPPCState *env)
   {
       /* Breakpoints */
@@ -6435,6 +6449,7 @@ static void init_proc_POWER10(CPUPPCState *env)
       /* Common Registers */
       init_proc_book3s_common(env);
       register_book3s_207_dbg_sprs(env);
+    register_book3s_310_dbg_sprs(env);

       /* Common TCG PMU */
       init_tcg_pmu_power8(env);
diff --git a/target/ppc/kvm.c b/target/ppc/kvm.c
index a8a935e267..05f68d2d91 100644
--- a/target/ppc/kvm.c
+++ b/target/ppc/kvm.c
@@ -89,6 +89,7 @@ static int cap_large_decr;
   static int cap_fwnmi;
   static int cap_rpt_invalidate;
   static int cap_ail_mode_3;
+static int cap_dawr1;

   static uint32_t debug_inst_opcode;

@@ -138,6 +139,7 @@ int kvm_arch_init(MachineState *ms, KVMState *s)
       cap_ppc_nested_kvm_hv = kvm_vm_check_extension(s, KVM_CAP_PPC_NESTED_HV);
       cap_large_decr = kvmppc_get_dec_bits();
       cap_fwnmi = kvm_vm_check_extension(s, KVM_CAP_PPC_FWNMI);
+    cap_dawr1 = kvm_vm_check_extension(s, KVM_CAP_PPC_DAWR1);
       /*
        * Note: setting it to false because there is not such capability
        * in KVM at this moment.
@@ -2109,6 +2111,16 @@ int kvmppc_set_fwnmi(PowerPCCPU *cpu)
       return kvm_vcpu_enable_cap(cs, KVM_CAP_PPC_FWNMI, 0);
   }

+bool kvmppc_has_cap_dawr1(void)
+{
+    return !!cap_dawr1;
+}
+
+int kvmppc_set_cap_dawr1(int enable)
+{
+    return kvm_vm_enable_cap(kvm_state, KVM_CAP_PPC_DAWR1, 0, enable);
+}
+
   int kvmppc_smt_threads(void)
   {
       return cap_ppc_smt ? cap_ppc_smt : 1;
diff --git a/target/ppc/kvm_ppc.h b/target/ppc/kvm_ppc.h
index 611debc3ce..584916a6d1 100644
--- a/target/ppc/kvm_ppc.h
+++ b/target/ppc/kvm_ppc.h
@@ -67,6 +67,8 @@ bool kvmppc_has_cap_htm(void);
   bool kvmppc_has_cap_mmu_radix(void);
   bool kvmppc_has_cap_mmu_hash_v3(void);
   bool kvmppc_has_cap_xive(void);
+bool kvmppc_has_cap_dawr1(void);
+int kvmppc_set_cap_dawr1(int enable);
   int kvmppc_get_cap_safe_cache(void);
   int kvmppc_get_cap_safe_bounds_check(void);
   int kvmppc_get_cap_safe_indirect_branch(void);
@@ -352,6 +354,16 @@ static inline bool kvmppc_has_cap_xive(void)
       return false;
   }

+static inline bool kvmppc_has_cap_dawr1(void)
+{
+    return false;
+}
+
+static inline int kvmppc_set_cap_dawr1(int enable)
+{
+    abort();
+}
+
   static inline int kvmppc_get_cap_safe_cache(void)
   {
       return 0;









reply via email to

[Prev in Thread] Current Thread [Next in Thread]