Skip to contentSkip to conversation

qxl: remove assert and migration blocker

PATCH In review Gerd Hoffmann 5 participants 13 messages Jul 21, 2021 – Jul 23, 2021

lore
Share & tools
lore
#0 lore raw

cover qxl: remove assert and migration blocker

Gerd Hoffmann (2):
qxl: remove assert in qxl_pre_save.
Revert "qxl: add migration blocker to avoid pre-save assert"

 hw/display/qxl.h |  1 -
 hw/display/qxl.c | 35 +++--------------------------------
 2 files changed, 3 insertions(+), 33 deletions(-)
Signature · 1 line
2.31.1
#1 lore raw

PATCH 1/2 qxl: remove assert in qxl_pre_save.

Since commit 551dbd0846d2 ("migration: check pre_save return in vmstate_save_state") the pre_save hook can fail. So lets finally use that to drop the guest-triggerable assert in qxl_pre_save().

---
 hw/display/qxl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
hw/display/qxl.c +3 -1
index 84f99088e0a0..3867b94fe236 100644
@@ -2283,7 +2283,9 @@ static int qxl_pre_save(void *opaque)
     } else {
         d->last_release_offset = (uint8_t *)d->last_release - ram_start;
     }
-    assert(d->last_release_offset < d->vga.vram_size);
+    if (d->last_release_offset < d->vga.vram_size) {
+        return 1;
+    }
 
     return 0;
 }
Signature · 1 line
2.31.1
Reviewed-by on v1 1/2
#2 lore raw

Re: [PATCH 1/2] qxl: remove assert in qxl_pre_save.

Quoted 6 lines · Gerd Hoffmann

Since commit 551dbd0846d2 ("migration: check pre_save return in vmstate_save_state") the pre_save hook can fail. So lets finally use that to drop the guest-triggerable assert in qxl_pre_save().

---

Quoted 23 lines
 hw/display/qxl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
hw/display/qxl.c +3 -2
index 84f99088e0a0..3867b94fe236 100644
@@ -2283,7 +2283,9 @@ static int qxl_pre_save(void *opaque)
     } else {
         d->last_release_offset = (uint8_t *)d->last_release - ram_start;
     }
-    assert(d->last_release_offset < d->vga.vram_size);
+    if (d->last_release_offset < d->vga.vram_size) {
+        return 1;
+    }
 
     return 0;
 }
--

2.31.1

Signature · 1 line
Marc-André Lureau
#3 lore raw

Re: [PATCH 1/2] qxl: remove assert in qxl_pre_save.

In reply to Gerd Hoffmann
Quoted 14 lines · * Gerd Hoffmann ([email protected]) wrote:

Since commit 551dbd0846d2 ("migration: check pre_save return in vmstate_save_state") the pre_save hook can fail. So lets finally use that to drop the guest-triggerable assert in qxl_pre_save().

---
 hw/display/qxl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
hw/display/qxl.c +0 -0
index 84f99088e0a0..3867b94fe236 100644

hw/display/qxl.c static int qxl_pre_save(void *opaque) · line 2287

Quoted 6 lines
hw/display/qxl.c +2 -1
@@ -2283,7 +2283,9 @@ static int qxl_pre_save(void *opaque)
     } else {
         d->last_release_offset = (uint8_t *)d->last_release - ram_start;
     }
-    assert(d->last_release_offset < d->vga.vram_size);
+    if (d->last_release_offset < d->vga.vram_size) {
+        return 1;

It would be great if there was an error_report or something there so that we get some idea of what happened.

Dave

Quoted 8 lines
hw/display/qxl.c +1 -0
+    }
 
     return 0;
 }
Signature · 1 line
2.31.1
Signature · 1 line
Dr. David Alan Gilbert / [email protected] / Manchester, UK
#4 lore raw

Re: [PATCH 1/2] qxl: remove assert in qxl_pre_save.

Quoted 10 lines · Dr. David Alan Gilbert
Quoted 9 lines · * Gerd Hoffmann ([email protected]) wrote:

Since commit 551dbd0846d2 ("migration: check pre_save return in vmstate_save_state") the pre_save hook can fail. So lets finally use that to drop the guest-triggerable assert in qxl_pre_save().

---
 hw/display/qxl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Quoted 14 lines

hw/display/qxl.c static int qxl_pre_save(void *opaque) · line 2287

Quoted 11 lines
hw/display/qxl.c +2 -1
index 84f99088e0a0..3867b94fe236 100644
@@ -2283,7 +2283,9 @@ static int qxl_pre_save(void *opaque)
     } else {
         d->last_release_offset = (uint8_t *)d->last_release - ram_start;
     }
-    assert(d->last_release_offset < d->vga.vram_size);
+    if (d->last_release_offset < d->vga.vram_size) {
+        return 1;

It would be great if there was an error_report or something there so that we get some idea of what happened.

Hmm. Strictly speaking qemu_log_mask(LOG_GUEST_ERROR, "..."); would be the correct approach here, but I guess that wouldn't help much for trouble-shooting as the message would only show up with "-d guest_errors". So better make that error_report("...") ?

take care,
Gerd

#5 lore raw

Re: [PATCH 1/2] qxl: remove assert in qxl_pre_save.

Quoted 26 lines · Gerd Hoffmann
Quoted 10 lines · Dr. David Alan Gilbert
Quoted 9 lines · * Gerd Hoffmann ([email protected]) wrote:

Since commit 551dbd0846d2 ("migration: check pre_save return in vmstate_save_state") the pre_save hook can fail. So lets finally use that to drop the guest-triggerable assert in qxl_pre_save().

---
 hw/display/qxl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Quoted 14 lines

hw/display/qxl.c static int qxl_pre_save(void *opaque) · line 2287

Quoted 11 lines
hw/display/qxl.c +2 -1
index 84f99088e0a0..3867b94fe236 100644
@@ -2283,7 +2283,9 @@ static int qxl_pre_save(void *opaque)
     } else {
         d->last_release_offset = (uint8_t *)d->last_release - ram_start;
     }
-    assert(d->last_release_offset < d->vga.vram_size);
+    if (d->last_release_offset < d->vga.vram_size) {
+        return 1;

It would be great if there was an error_report or something there so that we get some idea of what happened.

Quoted 4 lines

Hmm. Strictly speaking qemu_log_mask(LOG_GUEST_ERROR, "..."); would be the correct approach here, but I guess that wouldn't help much for trouble-shooting as the message would only show up with "-d guest_errors". So better make that error_report("...") ?

It depends who is interested in such errors. Does management app care?

#6 lore raw

Re: [PATCH 1/2] qxl: remove assert in qxl_pre_save.

Quoted 27 lines · Philippe Mathieu-Daudé
Quoted 26 lines · Gerd Hoffmann
Quoted 24 lines · Dr. David Alan Gilbert
Quoted 14 lines · * Gerd Hoffmann ([email protected]) wrote:

Since commit 551dbd0846d2 ("migration: check pre_save return in vmstate_save_state") the pre_save hook can fail. So lets finally use that to drop the guest-triggerable assert in qxl_pre_save().

---
 hw/display/qxl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
hw/display/qxl.c +0 -0
index 84f99088e0a0..3867b94fe236 100644

hw/display/qxl.c static int qxl_pre_save(void *opaque) · line 2287

Quoted 6 lines
hw/display/qxl.c +2 -1
@@ -2283,7 +2283,9 @@ static int qxl_pre_save(void *opaque)
     } else {
         d->last_release_offset = (uint8_t *)d->last_release - ram_start;
     }
-    assert(d->last_release_offset < d->vga.vram_size);
+    if (d->last_release_offset < d->vga.vram_size) {
+        return 1;

It would be great if there was an error_report or something there so that we get some idea of what happened.

Quoted 6 lines
Quoted 4 lines

Hmm. Strictly speaking qemu_log_mask(LOG_GUEST_ERROR, "..."); would be the correct approach here, but I guess that wouldn't help much for trouble-shooting as the message would only show up with "-d guest_errors". So better make that error_report("...") ?

It depends who is interested in such errors.

Migration fails and you want know why.

Reason migration fails is the guest violates qxl specs (which doesn't cause actual problems unless you want migrate the guest).

take care,
Gerd

#7 lore raw

Re: [PATCH 1/2] qxl: remove assert in qxl_pre_save.

Quoted 33 lines · Gerd Hoffmann
Quoted 32 lines · Philippe Mathieu-Daudé
Quoted 30 lines · Gerd Hoffmann
Quoted 10 lines · Dr. David Alan Gilbert
Quoted 9 lines · * Gerd Hoffmann ([email protected]) wrote:

Since commit 551dbd0846d2 ("migration: check pre_save return in vmstate_save_state") the pre_save hook can fail. So lets finally use that to drop the guest-triggerable assert in qxl_pre_save().

---
 hw/display/qxl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Quoted 14 lines

hw/display/qxl.c static int qxl_pre_save(void *opaque) · line 2287

Quoted 11 lines
hw/display/qxl.c +2 -1
index 84f99088e0a0..3867b94fe236 100644
@@ -2283,7 +2283,9 @@ static int qxl_pre_save(void *opaque)
     } else {
         d->last_release_offset = (uint8_t *)d->last_release - ram_start;
     }
-    assert(d->last_release_offset < d->vga.vram_size);
+    if (d->last_release_offset < d->vga.vram_size) {
+        return 1;

It would be great if there was an error_report or something there so that we get some idea of what happened.

Hmm. Strictly speaking qemu_log_mask(LOG_GUEST_ERROR, "..."); would be the correct approach here, but I guess that wouldn't help much for trouble-shooting as the message would only show up with "-d guest_errors". So better make that error_report("...") ?

Quoted 6 lines
Quoted 1 line

It depends who is interested in such errors.

Migration fails and you want know why.

Reason migration fails is the guest violates qxl specs (which doesn't cause actual problems unless you want migrate the guest).

Then I agree error_report() is better ("Report an error to the current monitor if we have one, else stderr").

#8 lore raw

Re: [PATCH 1/2] qxl: remove assert in qxl_pre_save.

Quoted 36 lines · Philippe Mathieu-Daudé
Quoted 35 lines · Gerd Hoffmann
Quoted 33 lines · Philippe Mathieu-Daudé
Quoted 26 lines · Gerd Hoffmann
Quoted 10 lines · Dr. David Alan Gilbert
Quoted 9 lines · * Gerd Hoffmann ([email protected]) wrote:

Since commit 551dbd0846d2 ("migration: check pre_save return in vmstate_save_state") the pre_save hook can fail. So lets finally use that to drop the guest-triggerable assert in qxl_pre_save().

---
 hw/display/qxl.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
Quoted 14 lines

hw/display/qxl.c static int qxl_pre_save(void *opaque) · line 2287

Quoted 11 lines
hw/display/qxl.c +2 -1
index 84f99088e0a0..3867b94fe236 100644
@@ -2283,7 +2283,9 @@ static int qxl_pre_save(void *opaque)
     } else {
         d->last_release_offset = (uint8_t *)d->last_release - ram_start;
     }
-    assert(d->last_release_offset < d->vga.vram_size);
+    if (d->last_release_offset < d->vga.vram_size) {
+        return 1;

It would be great if there was an error_report or something there so that we get some idea of what happened.

Quoted 4 lines

Hmm. Strictly speaking qemu_log_mask(LOG_GUEST_ERROR, "..."); would be the correct approach here, but I guess that wouldn't help much for trouble-shooting as the message would only show up with "-d guest_errors". So better make that error_report("...") ?

It depends who is interested in such errors.

Quoted 7 lines
Quoted 4 lines

Migration fails and you want know why.

Reason migration fails is the guest violates qxl specs (which doesn't cause actual problems unless you want migrate the guest).

Then I agree error_report() is better ("Report an error to the current monitor if we have one, else stderr").

No good for QMP, I'm afraid.

error_report() reports to stderr in QMP monitor context. To get an error to the QMP client, you need to use error_setg() and propagate all the way to the QMP core.

Back to the patch. If the thing we report is actually an *error*, i.e. something that makes migration fail, then the QMP failure will contain an error message. That error must come from further up the call stack. If it's good enough, we don't need error_report() here. Else, we should error_setg() here.

If it's not actually an error, then error_report() is wrong. warn_report() might be right.

#9 lore raw

Re: [PATCH 1/2] qxl: remove assert in qxl_pre_save.

Hi,

Quoted 14 lines
Quoted 7 lines
Quoted 4 lines

Migration fails and you want know why.

Reason migration fails is the guest violates qxl specs (which doesn't cause actual problems unless you want migrate the guest).

Then I agree error_report() is better ("Report an error to the current monitor if we have one, else stderr").

No good for QMP, I'm afraid.

error_report() reports to stderr in QMP monitor context. To get an error to the QMP client, you need to use error_setg() and propagate all the way to the QMP core.

Quoted 5 lines

Back to the patch. If the thing we report is actually an *error*, i.e. something that makes migration fail, then the QMP failure will contain an error message. That error must come from further up the call stack. If it's good enough, we don't need error_report() here. Else, we should error_setg() here.

Caller is here (migration/vmstate.c):

int vmstate_save_state_v(QEMUFile *f, const VMStateDescription *vmsd,
void *opaque, JSONWriter *vmdesc, int version_id)

{
    [ ... ]
    if (vmsd->pre_save) {
        ret = vmsd->pre_save(opaque);
        trace_vmstate_save_state_pre_save_res(vmsd->name, ret);
        if (ret) {
            error_report("pre-save failed: %s", vmsd->name);
            return ret;
        }
    }
    [ ... ]
}

No error_setg() anywhere close, it's error_report() all over the place. I guess that makes adding error_report() the only short-term (aka 6.1) option.

take care,
Gerd

#10 lore raw

PATCH 2/2 Revert "qxl: add migration blocker to avoid pre-save assert"

In reply to Gerd Hoffmann

This reverts commit 86dbcdd9c7590d06db89ca256c5eaf0b4aba8858.

The pre-save assert is gone now, so the migration blocker is not needed any more.

---
 hw/display/qxl.h |  1 -
 hw/display/qxl.c | 31 -------------------------------
 2 files changed, 32 deletions(-)
hw/display/qxl.h +0 -1
index 379d3304abc1..30d21f4d0bdc 100644
@@ -39,7 +39,6 @@ struct PCIQXLDevice {
     uint32_t           cmdlog;
 
     uint32_t           guest_bug;
-    Error              *migration_blocker;
 
     enum qxl_mode      mode;
     uint32_t           cmdflags;
hw/display/qxl.c +0 -31
index 3867b94fe236..43482d4364ba 100644
@@ -30,7 +30,6 @@
 #include "qemu/module.h"
 #include "hw/qdev-properties.h"
 #include "sysemu/runstate.h"
-#include "migration/blocker.h"
 #include "migration/vmstate.h"
 #include "trace.h"
 
@@ -666,30 +665,6 @@ static int interface_get_command(QXLInstance *sin, struct QXLCommandExt *ext)
         qxl->guest_primary.commands++;
         qxl_track_command(qxl, ext);
         qxl_log_command(qxl, "cmd", ext);
-        {
-            /*
-             * Windows 8 drivers place qxl commands in the vram
-             * (instead of the ram) bar.  We can't live migrate such a
-             * guest, so add a migration blocker in case we detect
-             * this, to avoid triggering the assert in pre_save().
-             *
-             * https://cgit.freedesktop.org/spice/win32/qxl-wddm-dod/commit/?id=f6e099db39e7d0787f294d5fd0dce328b5210faa
-             */
-            void *msg = qxl_phys2virt(qxl, ext->cmd.data, ext->group_id);
-            if (msg != NULL && (
-                    msg < (void *)qxl->vga.vram_ptr ||
-                    msg > ((void *)qxl->vga.vram_ptr + qxl->vga.vram_size))) {
-                if (!qxl->migration_blocker) {
-                    Error *local_err = NULL;
-                    error_setg(&qxl->migration_blocker,
-                               "qxl: guest bug: command not in ram bar");
-                    migrate_add_blocker(qxl->migration_blocker, &local_err);
-                    if (local_err) {
-                        error_report_err(local_err);
-                    }
-                }
-            }
-        }
         trace_qxl_ring_command_get(qxl->id, qxl_mode_to_string(qxl->mode));
         return true;
     default:
@@ -1283,12 +1258,6 @@ static void qxl_hard_reset(PCIQXLDevice *d, int loadvm)
     qemu_spice_create_host_memslot(&d->ssd);
     qxl_soft_reset(d);
 
-    if (d->migration_blocker) {
-        migrate_del_blocker(d->migration_blocker);
-        error_free(d->migration_blocker);
-        d->migration_blocker = NULL;
-    }
-
     if (startstop) {
         qemu_spice_display_start();
     }
Signature · 1 line
2.31.1
Reviewed-by on v1 2/2
#11 lore raw

Re: [PATCH 2/2] Revert "qxl: add migration blocker to avoid pre-save assert"

Quoted 7 lines · Gerd Hoffmann

This reverts commit 86dbcdd9c7590d06db89ca256c5eaf0b4aba8858.

The pre-save assert is gone now, so the migration blocker is not needed any more.

Quoted 84 lines
---
 hw/display/qxl.h |  1 -
 hw/display/qxl.c | 31 -------------------------------
 2 files changed, 32 deletions(-)
hw/display/qxl.h +0 -1
index 379d3304abc1..30d21f4d0bdc 100644
@@ -39,7 +39,6 @@ struct PCIQXLDevice {
     uint32_t           cmdlog;
 
     uint32_t           guest_bug;
-    Error              *migration_blocker;
 
     enum qxl_mode      mode;
     uint32_t           cmdflags;
hw/display/qxl.c +0 -9
index 3867b94fe236..43482d4364ba 100644
@@ -30,7 +30,6 @@
 #include "qemu/module.h"
 #include "hw/qdev-properties.h"
 #include "sysemu/runstate.h"
-#include "migration/blocker.h"
 #include "migration/vmstate.h"
 #include "trace.h"
 
@@ -666,30 +665,6 @@ static int interface_get_command(QXLInstance *sin,
 struct QXLCommandExt *ext)
         qxl->guest_primary.commands++;
         qxl_track_command(qxl, ext);
         qxl_log_command(qxl, "cmd", ext);
-        {
-            /*
-             * Windows 8 drivers place qxl commands in the vram
-             * (instead of the ram) bar.  We can't live migrate such a
-             * guest, so add a migration blocker in case we detect
-             * this, to avoid triggering the assert in pre_save().
-             *
-             *

https://cgit.freedesktop.org/spice/win32/qxl-wddm-dod/commit/?id=f6e099db39e7d0787f294d5fd0dce328b5210faa

hw/display/qxl.c +0 -5
-             */
-            void *msg = qxl_phys2virt(qxl, ext->cmd.data, ext->group_id);
-            if (msg != NULL && (
-                    msg < (void *)qxl->vga.vram_ptr ||
-                    msg > ((void *)qxl->vga.vram_ptr +

qxl->vga.vram_size))) {

hw/display/qxl.c +0 -5
-                if (!qxl->migration_blocker) {
-                    Error *local_err = NULL;
-                    error_setg(&qxl->migration_blocker,
-                               "qxl: guest bug: command not in ram bar");
-                    migrate_add_blocker(qxl->migration_blocker,

&local_err);

hw/display/qxl.c +0 -6
-                    if (local_err) {
-                        error_report_err(local_err);
-                    }
-                }
-            }
-        }
         trace_qxl_ring_command_get(qxl->id,
qxl_mode_to_string(qxl->mode));
         return true;
     default:
hw/display/qxl.c +0 -6
@@ -1283,12 +1258,6 @@ static void qxl_hard_reset(PCIQXLDevice *d, int
 loadvm)
     qemu_spice_create_host_memslot(&d->ssd);
     qxl_soft_reset(d);
 
-    if (d->migration_blocker) {
-        migrate_del_blocker(d->migration_blocker);
-        error_free(d->migration_blocker);
-        d->migration_blocker = NULL;
-    }
-
     if (startstop) {
         qemu_spice_display_start();
     }

--
2.31.1

Signature · 1 line
Marc-André Lureau
#12 lore raw

Re: [PATCH 2/2] Revert "qxl: add migration blocker to avoid pre-save assert"

In reply to Gerd Hoffmann
Quoted 56 lines · Gerd Hoffmann

This reverts commit 86dbcdd9c7590d06db89ca256c5eaf0b4aba8858.

The pre-save assert is gone now, so the migration blocker is not needed any more.

---
 hw/display/qxl.h |  1 -
 hw/display/qxl.c | 31 -------------------------------
 2 files changed, 32 deletions(-)
hw/display/qxl.h +0 -1
index 379d3304abc1..30d21f4d0bdc 100644
@@ -39,7 +39,6 @@ struct PCIQXLDevice {
     uint32_t           cmdlog;
 
     uint32_t           guest_bug;
-    Error              *migration_blocker;
 
     enum qxl_mode      mode;
     uint32_t           cmdflags;
hw/display/qxl.c +0 -18
index 3867b94fe236..43482d4364ba 100644
@@ -30,7 +30,6 @@
 #include "qemu/module.h"
 #include "hw/qdev-properties.h"
 #include "sysemu/runstate.h"
-#include "migration/blocker.h"
 #include "migration/vmstate.h"
 #include "trace.h"
 
@@ -666,30 +665,6 @@ static int interface_get_command(QXLInstance *sin, struct QXLCommandExt *ext)
         qxl->guest_primary.commands++;
         qxl_track_command(qxl, ext);
         qxl_log_command(qxl, "cmd", ext);
-        {
-            /*
-             * Windows 8 drivers place qxl commands in the vram
-             * (instead of the ram) bar.  We can't live migrate such a
-             * guest, so add a migration blocker in case we detect
-             * this, to avoid triggering the assert in pre_save().
-             *
-             * https://cgit.freedesktop.org/spice/win32/qxl-wddm-dod/commit/?id=f6e099db39e7d0787f294d5fd0dce328b5210faa
-             */
-            void *msg = qxl_phys2virt(qxl, ext->cmd.data, ext->group_id);
-            if (msg != NULL && (
-                    msg < (void *)qxl->vga.vram_ptr ||
-                    msg > ((void *)qxl->vga.vram_ptr + qxl->vga.vram_size))) {
-                if (!qxl->migration_blocker) {
-                    Error *local_err = NULL;
-                    error_setg(&qxl->migration_blocker,
-                               "qxl: guest bug: command not in ram bar");

hw/display/qxl.c static int interface_get_command(QXLInstance *sin, struct QXLCommandExt *ext) · line 670

Quoted 10 lines
hw/display/qxl.c +0 -7
-                    migrate_add_blocker(qxl->migration_blocker, &local_err);
-                    if (local_err) {
-                        error_report_err(local_err);
-                    }
-                }
-            }
-        }
         trace_qxl_ring_command_get(qxl->id, qxl_mode_to_string(qxl->mode));
         return true;
     default:

Conflicts with my "[PATCH 09/16] migration: Unify failure check for migrate_add_blocker()". Easy to resolve: your patch goes first, I drop the hunk from mine.

Quoted 13 lines
hw/display/qxl.c +0 -6
@@ -1283,12 +1258,6 @@ static void qxl_hard_reset(PCIQXLDevice *d, int loadvm)
     qemu_spice_create_host_memslot(&d->ssd);
     qxl_soft_reset(d);
 
-    if (d->migration_blocker) {
-        migrate_del_blocker(d->migration_blocker);
-        error_free(d->migration_blocker);
-        d->migration_blocker = NULL;
-    }
-
     if (startstop) {
         qemu_spice_display_start();
     }