Skip to contentSkip to conversation

memory: Do not print MR priority in flatview HMP output

PATCH In review Philippe Mathieu-Daudé 3 participants 4 messages Dec 28, 2022 – Jan 3, 2023

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 memory: Do not print MR priority in flatview HMP output

When requesting the flatview output with 'info mtree -f', the MemoryRegion priority is irrelevant and noise. Remove it.

---
 softmmu/memory.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
softmmu/memory.c +2 -4
index e05332d07f..89713dd5ce 100644
@@ -3330,22 +3330,20 @@ static void mtree_print_flatview(gpointer key, gpointer value,
         mr = range->mr;
         if (range->offset_in_region) {
             qemu_printf(MTREE_INDENT TARGET_FMT_plx "-" TARGET_FMT_plx
-                        " (prio %d, %s%s): %s @" TARGET_FMT_plx,
+                        " (%s%s): %s @" TARGET_FMT_plx,
                         int128_get64(range->addr.start),
                         int128_get64(range->addr.start)
                         + MR_SIZE(range->addr.size),
-                        mr->priority,
                         range->nonvolatile ? "nv-" : "",
                         range->readonly ? "rom" : memory_region_type(mr),
                         memory_region_name(mr),
                         range->offset_in_region);
         } else {
             qemu_printf(MTREE_INDENT TARGET_FMT_plx "-" TARGET_FMT_plx
-                        " (prio %d, %s%s): %s",
+                        " (%s%s): %s",
                         int128_get64(range->addr.start),
                         int128_get64(range->addr.start)
                         + MR_SIZE(range->addr.size),
-                        mr->priority,
                         range->nonvolatile ? "nv-" : "",
                         range->readonly ? "rom" : memory_region_type(mr),
                         memory_region_name(mr));
Signature · 1 line
2.38.1
#1 lore raw

Is there any chance between 7.1 and 7.2 ‘something’ happened to make it so that Qemu ‘cares more’ about e.g. when memory regions are added/removed?

I seem to get an abort because a memory region has not been completely setup in 7.2 (while it is being flattened actually) - In 7.1 that never seemed to happen….? (Putting it all async work seems ‘excessive’ but maybe it’s needed now?)

Cheers
Mark

On 28/12/2022, 14:04, "Philippe Mathieu-Daudé" <[email protected]> wrote:

WARNING: This email originated from outside of Qualcomm. Please be wary of any links or attachments, and do not enable macros.

When requesting the flatview output with 'info mtree -f', the MemoryRegion priority is irrelevant and noise. Remove it.

---
 softmmu/memory.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)
softmmu/memory.c +2 -4
index e05332d07f..89713dd5ce 100644
@@ -3330,22 +3330,20 @@ static void mtree_print_flatview(gpointer key, gpointer value,
         mr = range->mr;
         if (range->offset_in_region) {
             qemu_printf(MTREE_INDENT TARGET_FMT_plx "-" TARGET_FMT_plx
-                        " (prio %d, %s%s): %s @" TARGET_FMT_plx,
+                        " (%s%s): %s @" TARGET_FMT_plx,
                         int128_get64(range->addr.start),
                         int128_get64(range->addr.start)
                         + MR_SIZE(range->addr.size),
-                        mr->priority,
                         range->nonvolatile ? "nv-" : "",
                         range->readonly ? "rom" : memory_region_type(mr),
                         memory_region_name(mr),
                         range->offset_in_region);
         } else {
             qemu_printf(MTREE_INDENT TARGET_FMT_plx "-" TARGET_FMT_plx
-                        " (prio %d, %s%s): %s",
+                        " (%s%s): %s",
                         int128_get64(range->addr.start),
                         int128_get64(range->addr.start)
                         + MR_SIZE(range->addr.size),
-                        mr->priority,
                         range->nonvolatile ? "nv-" : "",
                         range->readonly ? "rom" : memory_region_type(mr),
                         memory_region_name(mr));

--
2.38.1

#2 lore raw
Quoted 3 lines · Mark Burton

Is there any chance between 7.1 and 7.2 ‘something’ happened to make it so that Qemu ‘cares more’ about e.g. when memory regions are added/removed?

I seem to get an abort because a memory region has not been completely setup in 7.2 (while it is being flattened actually) - In 7.1 that never seemed to happen….?

I couldn't find any relevant changes in softmmu/. Maybe TCG related?

$ git diff --stat v7.1.0..v7.2.0 -- softmmu/
 softmmu/cpus.c         | 18 +++++++++++++-----
 softmmu/device_tree.c  | 58 
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 softmmu/dirtylimit.c   |  4 ++--
 softmmu/main.c         | 10 +++++-----
 softmmu/memory.c       | 72 
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 softmmu/physmem.c      | 36 ++++++++++++++++++------------------
 softmmu/qemu-seccomp.c | 13 +++++++++++++
 softmmu/qtest.c        | 12 +++++++++---
 softmmu/runstate.c     | 11 ++++++++---
 softmmu/vl.c           | 85 
+++++++++++++++++++++++++++++++++++++++++++++++++++----------------------------------
 10 files changed, 249 insertions(+), 70 deletions(-)
$ git diff --stat v7.1.0..v7.2.0 -- accel/tcg/
 accel/tcg/cpu-exec-common.c     |    2 +-
 accel/tcg/cpu-exec.c            |  229 +++++++++++++++------------
 accel/tcg/cputlb.c              |  350 
+++++++++++++++++++++---------------------
 accel/tcg/internal.h            |  101 ++++++++++++
 accel/tcg/meson.build           |    1 +
 accel/tcg/plugin-gen.c          |   22 +--
 accel/tcg/tb-hash.h             |    1 +
 accel/tcg/tb-jmp-cache.h        |   65 ++++++++
 accel/tcg/tb-maint.c            |  704 
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 accel/tcg/tcg-accel-ops-mttcg.c |    5 +-
 accel/tcg/tcg-accel-ops-rr.c    |    9 +-
 accel/tcg/tcg-accel-ops.c       |   98 ++++++++++++
 accel/tcg/tcg-all.c             |   17 +++
 accel/tcg/translate-all.c       | 1249 
+++++++++++++++++++---------------------------------------------------------------------------------------------------------------------------------
 accel/tcg/translator.c          |  138 ++++++++++++-----
 accel/tcg/user-exec.c           |   59 ++++++-
 16 files changed, 1628 insertions(+), 1422 deletions(-)

I'll have a look at

  $ git log -p v7.1.0..v7.2.0 accel/tcg/
Quoted 5 lines

(Putting it all async work seems ‘excessive’ but maybe it’s needed now?)

Cheers
Mark

#3 lore raw
In reply to Philippe Mathieu-Daudé
Quoted 2 lines · Philippe Mathieu-Daudé

When requesting the flatview output with 'info mtree -f', the MemoryRegion priority is irrelevant and noise. Remove it.

It is in many cases still helpful debugging information to me.

Maybe it's because it can cause mis-alignment when the priority >= 10? If that's the case, how about tabbing it properly instead of removing?

Thanks,

Signature · 1 line
Peter Xu