diff mbox

target-cris: update CPU state save/load to use VMStateDescription

Message ID 1432569199-24306-1-git-send-email-peter.maydell@linaro.org
State Superseded
Headers show

Commit Message

Peter Maydell May 25, 2015, 3:53 p.m. UTC
From: Juan Quintela <quintela@redhat.com>

Update the CRIS CPU state save/load to use a VMStateDescription struct
rather than cpu_save/cpu_load functions.

Have to define TLBSet struct.
Multidimensional arrays in C are a mess, just unroll them.

Signed-off-by: Juan Quintela <quintela@redhat.com>
[PMM:
 * expand commit message a little since it's no longer one patch in
   a 35-patch series
 * add header/copyright comment to machine.c; credited copyright is
   Red Hat and author is Juan, since this commit gives the file all-new
   contents; license is LGPL-2-or-later, to match other target-cris code
 * remove hardcoded tab
 * add fields for locked_irq, interrupt_vector, fault_vector, trap_vector
 * drop minimum_version_id_old fields
 * bump version_id to 2 as we are not compatible with old state format
 * remove unnecessary hw/boards.h include]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
This is a patch of Juan's from way back in 2012, which I am resurrecting
because we now have only two CPUs which still use old-style non-VMState
save/load (CRIS and SPARC). If we can update them both we can drop the
machinery in the common code which supports this.

Notes:
 * CPUCRISState indent style is somewhat mismatched (cf load_info);
   I took the "minimal make checkpatch happy" approach, but am
   happy to do something else with the line changed here
 * I have added a copyright header to target-cris/machine.c, because it
   did not have one at all, and this commit effectively gives the
   file all-new contents. I have set it up as LGPLv2-or-later,
   copyright Red Hat, author Juan. Please let me know if you would
   prefer something else!
 * I added vmstate entries for four fields which did not previously
   get saved and restored, which is presumably a bug fix...
 * vmsave/vmload on the axis-dev88 board does not currently seem to
   work (among other obvious problems, there is no vmstate support
   in the interrupt controller), so we're limited to "looks good
   on code review" here.

 target-cris/cpu.h     |  13 ++--
 target-cris/machine.c | 160 +++++++++++++++++++++++---------------------------
 2 files changed, 81 insertions(+), 92 deletions(-)

Comments

Peter Maydell May 25, 2015, 4:35 p.m. UTC | #1
On 25 May 2015 at 16:53, Peter Maydell <peter.maydell@linaro.org> wrote:
> From: Juan Quintela <quintela@redhat.com>
>
> Update the CRIS CPU state save/load to use a VMStateDescription struct
> rather than cpu_save/cpu_load functions.
>
> Have to define TLBSet struct.
> Multidimensional arrays in C are a mess, just unroll them.
>
> Signed-off-by: Juan Quintela <quintela@redhat.com>
> [PMM:
>  * expand commit message a little since it's no longer one patch in
>    a 35-patch series
>  * add header/copyright comment to machine.c; credited copyright is
>    Red Hat and author is Juan, since this commit gives the file all-new
>    contents; license is LGPL-2-or-later, to match other target-cris code
>  * remove hardcoded tab
>  * add fields for locked_irq, interrupt_vector, fault_vector, trap_vector
>  * drop minimum_version_id_old fields
>  * bump version_id to 2 as we are not compatible with old state format
>  * remove unnecessary hw/boards.h include]
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
> This is a patch of Juan's from way back in 2012, which I am resurrecting
> because we now have only two CPUs which still use old-style non-VMState
> save/load (CRIS and SPARC). If we can update them both we can drop the
> machinery in the common code which supports this.
>
> Notes:
>  * CPUCRISState indent style is somewhat mismatched (cf load_info);
>    I took the "minimal make checkpatch happy" approach, but am
>    happy to do something else with the line changed here
>  * I have added a copyright header to target-cris/machine.c, because it
>    did not have one at all, and this commit effectively gives the
>    file all-new contents. I have set it up as LGPLv2-or-later,
>    copyright Red Hat, author Juan. Please let me know if you would
>    prefer something else!
>  * I added vmstate entries for four fields which did not previously
>    get saved and restored, which is presumably a bug fix...
>  * vmsave/vmload on the axis-dev88 board does not currently seem to
>    work (among other obvious problems, there is no vmstate support
>    in the interrupt controller), so we're limited to "looks good
>    on code review" here.

Oops, this has a couple of issues I only noticed when I started
looking at the SPARC vmstate:
 * forgot to register vmstate by setting cc->vmsd
 * vmstate should be of CRISCPU, not CPUCRISState

This is why it's a shame the board doesn't have vmsave/load support
for testing.

Will send out a v2.

-- PMM
Peter Maydell May 25, 2015, 6:30 p.m. UTC | #2
On 25 May 2015 at 17:35, Peter Maydell <peter.maydell@linaro.org> wrote:
> On 25 May 2015 at 16:53, Peter Maydell <peter.maydell@linaro.org> wrote:
>> From: Juan Quintela <quintela@redhat.com>
>>
>> Update the CRIS CPU state save/load to use a VMStateDescription struct
>> rather than cpu_save/cpu_load functions.

> Oops, this has a couple of issues I only noticed when I started
> looking at the SPARC vmstate:
>  * forgot to register vmstate by setting cc->vmsd
>  * vmstate should be of CRISCPU, not CPUCRISState

In looking at this I found that we currently have:
CPUs that set cc->vmsd: arm, i386, lm32, mips, moxie, ppc, s390x
CPUs that set dc->vmsd: alpha, m68k, microblaze, openrisc, sh4,
unicore32, xtensa
...an exactly even split.

Which of these is the recommended approach for new conversions?
CCing Andreas since this is a QOM CPU question...

thanks
-- PMM
Peter Maydell May 25, 2015, 6:50 p.m. UTC | #3
On 25 May 2015 at 19:46, Peter Crosthwaite <peter.crosthwaite@xilinx.com> wrote:
> On Mon, May 25, 2015 at 11:30 AM, Peter Maydell
> <peter.maydell@linaro.org> wrote:
>> In looking at this I found that we currently have:
>> CPUs that set cc->vmsd: arm, i386, lm32, mips, moxie, ppc, s390x
>> CPUs that set dc->vmsd: alpha, m68k, microblaze, openrisc, sh4,
>> unicore32, xtensa
>> ...an exactly even split.
>>
>> Which of these is the recommended approach for new conversions?
>
> Should it be DC? CPU should not have specific support for something
> that already works for TYPE_DEVICE. This seems similar to
> SysBusDevice::init where we progressively pushed everything up to the
> higher level over time.

I'm tempted to agree with that, except that all of the CPU
families I'd trust as "major, actively maintained" are in
the "set cc->vmsd" camp...

-- PMM
Peter Maydell May 26, 2015, 7:12 a.m. UTC | #4
On 26 May 2015 at 00:58, Edgar E. Iglesias <edgar.iglesias@gmail.com> wrote:
> That's all OK I think. We've never used the save/restore/migration on
> CRIS so you won't be breaking anything for us...

I would actually recommend fixing it in the board/device models
just for debugging purposes. It's really handy to be able to snapshot
a system just before it fails, so that your reproduce case for a
bug is "run image from snapshot for half a second" rather than
"wait a minute or more for an OS to boot and then run some command".

-- PMM
diff mbox

Patch

diff --git a/target-cris/cpu.h b/target-cris/cpu.h
index 677b38c..826af97 100644
--- a/target-cris/cpu.h
+++ b/target-cris/cpu.h
@@ -108,6 +108,11 @@ 
 
 #define NB_MMU_MODES 2
 
+typedef struct {
+    uint32_t hi;
+    uint32_t lo;
+} TLBSet;
+
 typedef struct CPUCRISState {
 	uint32_t regs[16];
 	/* P0 - P15 are referred to as special registers in the docs.  */
@@ -161,11 +166,7 @@  typedef struct CPUCRISState {
 	 *
 	 * One for I and another for D.
 	 */
-	struct
-	{
-		uint32_t hi;
-		uint32_t lo;
-	} tlbsets[2][4][16];
+        TLBSet tlbsets[2][4][16];
 
 	CPU_COMMON
 
@@ -227,8 +228,6 @@  enum {
 #define cpu_gen_code cpu_cris_gen_code
 #define cpu_signal_handler cpu_cris_signal_handler
 
-#define CPU_SAVE_VERSION 1
-
 /* MMU modes definitions */
 #define MMU_MODE0_SUFFIX _kernel
 #define MMU_MODE1_SUFFIX _user
diff --git a/target-cris/machine.c b/target-cris/machine.c
index 8f9c0dd..8e65327 100644
--- a/target-cris/machine.c
+++ b/target-cris/machine.c
@@ -1,90 +1,80 @@ 
-#include "hw/hw.h"
-#include "hw/boards.h"
-
-void cpu_save(QEMUFile *f, void *opaque)
-{
-    CPUCRISState *env = opaque;
-    int i;
-    int s;
-    int mmu;
-
-    for (i = 0; i < 16; i++)
-        qemu_put_be32(f, env->regs[i]);
-    for (i = 0; i < 16; i++)
-        qemu_put_be32(f, env->pregs[i]);
-
-    qemu_put_be32(f, env->pc);
-    qemu_put_be32(f, env->ksp);
-
-    qemu_put_be32(f, env->dslot);
-    qemu_put_be32(f, env->btaken);
-    qemu_put_be32(f, env->btarget);
-
-    qemu_put_be32(f, env->cc_op);
-    qemu_put_be32(f, env->cc_mask);
-    qemu_put_be32(f, env->cc_dest);
-    qemu_put_be32(f, env->cc_src);
-    qemu_put_be32(f, env->cc_result);
-    qemu_put_be32(f, env->cc_size);
-    qemu_put_be32(f, env->cc_x);
-
-    for (s = 0; s < 4; s++) {
-        for (i = 0; i < 16; i++)
-            qemu_put_be32(f, env->sregs[s][i]);
-    }
+/*
+ *  CRIS virtual CPU state save/load support
+ *
+ *  Copyright (c) 2012 Red Hat, Inc.
+ *  Written by Juan Quintela <quintela@redhat.com>
+ *
+ * This library is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2 of the License, or (at your option) any later version.
+ *
+ * This library is distributed in the hope that 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 Lesser General Public
+ * License along with this library; if not, see <http://www.gnu.org/licenses/>.
+ */
 
-    qemu_put_be32(f, env->mmu_rand_lfsr);
-    for (mmu = 0; mmu < 2; mmu++) {
-        for (s = 0; s < 4; s++) {
-            for (i = 0; i < 16; i++) {
-                qemu_put_be32(f, env->tlbsets[mmu][s][i].lo);
-                qemu_put_be32(f, env->tlbsets[mmu][s][i].hi);
-            }
-        }
-    }
-}
-
-int cpu_load(QEMUFile *f, void *opaque, int version_id)
-{
-	CPUCRISState *env = opaque;
-    int i;
-    int s;
-    int mmu;
-
-    for (i = 0; i < 16; i++)
-        env->regs[i] = qemu_get_be32(f);
-    for (i = 0; i < 16; i++)
-        env->pregs[i] = qemu_get_be32(f);
-
-    env->pc = qemu_get_be32(f);
-    env->ksp = qemu_get_be32(f);
-
-    env->dslot = qemu_get_be32(f);
-    env->btaken = qemu_get_be32(f);
-    env->btarget = qemu_get_be32(f);
-
-    env->cc_op = qemu_get_be32(f);
-    env->cc_mask = qemu_get_be32(f);
-    env->cc_dest = qemu_get_be32(f);
-    env->cc_src = qemu_get_be32(f);
-    env->cc_result = qemu_get_be32(f);
-    env->cc_size = qemu_get_be32(f);
-    env->cc_x = qemu_get_be32(f);
+#include "hw/hw.h"
 
-    for (s = 0; s < 4; s++) {
-        for (i = 0; i < 16; i++)
-            env->sregs[s][i] = qemu_get_be32(f);
+static const VMStateDescription vmstate_tlbset = {
+    .name = "cpu/tlbset",
+    .version_id = 1,
+    .minimum_version_id = 1,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT32(lo, TLBSet),
+        VMSTATE_UINT32(hi, TLBSet),
+        VMSTATE_END_OF_LIST()
     }
+};
 
-    env->mmu_rand_lfsr = qemu_get_be32(f);
-    for (mmu = 0; mmu < 2; mmu++) {
-        for (s = 0; s < 4; s++) {
-            for (i = 0; i < 16; i++) {
-                env->tlbsets[mmu][s][i].lo = qemu_get_be32(f);
-                env->tlbsets[mmu][s][i].hi = qemu_get_be32(f);
-            }
-        }
+const VMStateDescription vmstate_cpu = {
+    .name = "cpu",
+    .version_id = 2,
+    .minimum_version_id = 2,
+    .fields = (VMStateField[]) {
+        VMSTATE_UINT32_ARRAY(regs, CPUCRISState, 16),
+        VMSTATE_UINT32_ARRAY(pregs, CPUCRISState, 16),
+        VMSTATE_UINT32(pc, CPUCRISState),
+        VMSTATE_UINT32(ksp, CPUCRISState),
+        VMSTATE_INT32(dslot, CPUCRISState),
+        VMSTATE_INT32(btaken, CPUCRISState),
+        VMSTATE_UINT32(btarget, CPUCRISState),
+        VMSTATE_UINT32(cc_op, CPUCRISState),
+        VMSTATE_UINT32(cc_mask, CPUCRISState),
+        VMSTATE_UINT32(cc_dest, CPUCRISState),
+        VMSTATE_UINT32(cc_src, CPUCRISState),
+        VMSTATE_UINT32(cc_result, CPUCRISState),
+        VMSTATE_INT32(cc_size, CPUCRISState),
+        VMSTATE_INT32(cc_x, CPUCRISState),
+        VMSTATE_INT32(locked_irq, CPUCRISState),
+        VMSTATE_INT32(interrupt_vector, CPUCRISState),
+        VMSTATE_INT32(fault_vector, CPUCRISState),
+        VMSTATE_INT32(trap_vector, CPUCRISState),
+        VMSTATE_UINT32_ARRAY(sregs[0], CPUCRISState, 16),
+        VMSTATE_UINT32_ARRAY(sregs[1], CPUCRISState, 16),
+        VMSTATE_UINT32_ARRAY(sregs[2], CPUCRISState, 16),
+        VMSTATE_UINT32_ARRAY(sregs[3], CPUCRISState, 16),
+        VMSTATE_UINT32(mmu_rand_lfsr, CPUCRISState),
+        VMSTATE_STRUCT_ARRAY(tlbsets[0][0], CPUCRISState, 16, 0,
+                             vmstate_tlbset, TLBSet),
+        VMSTATE_STRUCT_ARRAY(tlbsets[0][1], CPUCRISState, 16, 0,
+                             vmstate_tlbset, TLBSet),
+        VMSTATE_STRUCT_ARRAY(tlbsets[0][2], CPUCRISState, 16, 0,
+                             vmstate_tlbset, TLBSet),
+        VMSTATE_STRUCT_ARRAY(tlbsets[0][3], CPUCRISState, 16, 0,
+                             vmstate_tlbset, TLBSet),
+        VMSTATE_STRUCT_ARRAY(tlbsets[1][0], CPUCRISState, 16, 0,
+                             vmstate_tlbset, TLBSet),
+        VMSTATE_STRUCT_ARRAY(tlbsets[1][1], CPUCRISState, 16, 0,
+                             vmstate_tlbset, TLBSet),
+        VMSTATE_STRUCT_ARRAY(tlbsets[1][2], CPUCRISState, 16, 0,
+                             vmstate_tlbset, TLBSet),
+        VMSTATE_STRUCT_ARRAY(tlbsets[1][3], CPUCRISState, 16, 0,
+                             vmstate_tlbset, TLBSet),
+        VMSTATE_END_OF_LIST()
     }
-
-    return 0;
-}
+};