Skip to contentSkip to conversation

configure: disable MTTCG for MIPS guests

RFC PATCH In review Alex Bennée 3 participants 6 messages Mar 20, 2020

lore
Share & tools
lore
#0 lore raw

RFC PATCH 1/1 configure: disable MTTCG for MIPS guests

While debugging check-acceptance failures I found an instability in the mips64el test case. Briefly the test case:

  retry.py -n 100 -c -- ./mips64el-softmmu/qemu-system-mips64el \
    -display none -vga none -serial mon:stdio \
    -machine malta -kernel ./vmlinux-4.7.0-rc1.I6400 \
    -cpu I6400 -smp 8 -vga std \
    -append "printk.time=0 clocksource=GIC console=tty0 console=ttyS0 panic=-1" \
    --no-reboot

Reports about a 9% failure rate:

Results summary:
0: 91 times (91.00%), avg time 5.547 (0.45 varience/0.67 deviation) -6: 9 times (9.00%), avg time 3.394 (0.02 varience/0.13 deviation) Ran command 100 times, 91 passes

When re-run with "--accel tcg,thread=single" the instability goes away.

Results summary:
0: 100 times (100.00%), avg time 17.318 (249.76 varience/15.80 deviation) Ran command 100 times, 100 passes

Which seems to indicate there is some aspect of the MIPS MTTCG fixes that has been missed. Ideally we would fix that but I'm afraid I don't have time to investigate and am not super familiar with the architecture anyway.

I've disabled all the mips guests as I assume it's a fundamental synchronisation primitive that is broken but I haven't tested them all (there are a lot!).

Cc Aleksandar Markovic <[email protected]>
Cc Aurelien Jarno <[email protected]>
Cc Aleksandar Rikalo <[email protected]>
Cc Philippe Mathieu-Daudé <[email protected]>
---
 configure | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
configure +3 -3
index 206d22c5153..002792d21dc 100755
@@ -7832,19 +7832,19 @@ case "$target_name" in
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips|mipsel)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips
     echo "TARGET_ABI_MIPSO32=y" >> $config_target_mak
   ;;
   mipsn32|mipsn32el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN32=y" >> $config_target_mak
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips64|mips64el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN64=y" >> $config_target_mak
Signature · 1 line
2.20.1
Acked-by on v1 1/1
#1 lore raw
Quoted 13 lines · Alex Bennée

While debugging check-acceptance failures I found an instability in the mips64el test case. Briefly the test case:

  retry.py -n 100 -c -- ./mips64el-softmmu/qemu-system-mips64el \
    -display none -vga none -serial mon:stdio \
    -machine malta -kernel ./vmlinux-4.7.0-rc1.I6400 \
    -cpu I6400 -smp 8 -vga std \
    -append "printk.time=0 clocksource=GIC console=tty0 console=ttyS0 panic=-1" \
    --no-reboot

Reports about a 9% failure rate:

Quoted 7 lines

Results summary:
0: 91 times (91.00%), avg time 5.547 (0.45 varience/0.67 deviation) -6: 9 times (9.00%), avg time 3.394 (0.02 varience/0.13 deviation) Ran command 100 times, 91 passes

When re-run with "--accel tcg,thread=single" the instability goes away.

I confirm, in my notes I use this option with -smp with the following cpus:
- mips64dspr2
- I6400

Quoted 28 lines

Results summary:
0: 100 times (100.00%), avg time 17.318 (249.76 varience/15.80 deviation) Ran command 100 times, 100 passes

Which seems to indicate there is some aspect of the MIPS MTTCG fixes that has been missed. Ideally we would fix that but I'm afraid I don't have time to investigate and am not super familiar with the architecture anyway.

I've disabled all the mips guests as I assume it's a fundamental synchronisation primitive that is broken but I haven't tested them all (there are a lot!).

Cc Aleksandar Markovic <[email protected]>
Cc Aurelien Jarno <[email protected]>
Cc Aleksandar Rikalo <[email protected]>
Cc Philippe Mathieu-Daudé <[email protected]>
---
 configure | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
configure +0 -0
index 206d22c5153..002792d21dc 100755

configure case "$target_name" in · line 7835

Quoted 5 lines
configure +1 -1
@@ -7832,19 +7832,19 @@ case "$target_name" in
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips|mipsel)
-    mttcg="yes"
+    mttcg="no"

I am not sure the 32-bit targets are affected.
Maybe we can keep it enabled here?

For code below for 5.0 (64-bit targets):

Quoted 20 lines
configure +2 -3
     TARGET_ARCH=mips
     echo "TARGET_ABI_MIPSO32=y" >> $config_target_mak
   ;;
   mipsn32|mipsn32el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN32=y" >> $config_target_mak
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips64|mips64el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN64=y" >> $config_target_mak
--

2.20.1

#2 lore raw
In reply to Alex Bennée

пет, 20. мар 2020. у 12:45 Alex Bennée <[email protected]> је написао/ла:

Quoted 4 lines

While debugging check-acceptance failures I found an instability in the mips64el test case. Briefly the test case:

Alex, following up Philippe's hint, do you see failures while using 32-bit targets?

Thanks,
Aleksandar

Quoted 69 lines
  retry.py -n 100 -c -- ./mips64el-softmmu/qemu-system-mips64el \
    -display none -vga none -serial mon:stdio \
    -machine malta -kernel ./vmlinux-4.7.0-rc1.I6400 \
    -cpu I6400 -smp 8 -vga std \
    -append "printk.time=0 clocksource=GIC console=tty0 console=ttyS0 panic=-1" \
    --no-reboot

Reports about a 9% failure rate:

Results summary:
0: 91 times (91.00%), avg time 5.547 (0.45 varience/0.67 deviation) -6: 9 times (9.00%), avg time 3.394 (0.02 varience/0.13 deviation) Ran command 100 times, 91 passes

When re-run with "--accel tcg,thread=single" the instability goes away.

Results summary:
0: 100 times (100.00%), avg time 17.318 (249.76 varience/15.80 deviation) Ran command 100 times, 100 passes

Which seems to indicate there is some aspect of the MIPS MTTCG fixes that has been missed. Ideally we would fix that but I'm afraid I don't have time to investigate and am not super familiar with the architecture anyway.

I've disabled all the mips guests as I assume it's a fundamental synchronisation primitive that is broken but I haven't tested them all (there are a lot!).

Cc Aleksandar Markovic <[email protected]>
Cc Aurelien Jarno <[email protected]>
Cc Aleksandar Rikalo <[email protected]>
Cc Philippe Mathieu-Daudé <[email protected]>
---
 configure | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
configure +3 -4
index 206d22c5153..002792d21dc 100755
@@ -7832,19 +7832,19 @@ case "$target_name" in
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips|mipsel)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips
     echo "TARGET_ABI_MIPSO32=y" >> $config_target_mak
   ;;
   mipsn32|mipsn32el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN32=y" >> $config_target_mak
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips64|mips64el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN64=y" >> $config_target_mak
--

2.20.1

#3 lore raw
In reply to Alex Bennée

пет, 20. мар 2020. у 12:45 Alex Bennée <[email protected]> је написао/ла:

Quoted 11 lines

While debugging check-acceptance failures I found an instability in the mips64el test case. Briefly the test case:

  retry.py -n 100 -c -- ./mips64el-softmmu/qemu-system-mips64el \
    -display none -vga none -serial mon:stdio \
    -machine malta -kernel ./vmlinux-4.7.0-rc1.I6400 \
    -cpu I6400 -smp 8 -vga std \
    -append "printk.time=0 clocksource=GIC console=tty0 console=ttyS0 panic=-1" \
    --no-reboot

Thank for the findings!

Could you perhaps attach or link to "retry.py"?

Did you run this particular test for the first time now, or it used to pass before?

Thanks,
Aleksandar

Quoted 62 lines

Reports about a 9% failure rate:

Results summary:
0: 91 times (91.00%), avg time 5.547 (0.45 varience/0.67 deviation) -6: 9 times (9.00%), avg time 3.394 (0.02 varience/0.13 deviation) Ran command 100 times, 91 passes

When re-run with "--accel tcg,thread=single" the instability goes away.

Results summary:
0: 100 times (100.00%), avg time 17.318 (249.76 varience/15.80 deviation) Ran command 100 times, 100 passes

Which seems to indicate there is some aspect of the MIPS MTTCG fixes that has been missed. Ideally we would fix that but I'm afraid I don't have time to investigate and am not super familiar with the architecture anyway.

I've disabled all the mips guests as I assume it's a fundamental synchronisation primitive that is broken but I haven't tested them all (there are a lot!).

Cc Aleksandar Markovic <[email protected]>
Cc Aurelien Jarno <[email protected]>
Cc Aleksandar Rikalo <[email protected]>
Cc Philippe Mathieu-Daudé <[email protected]>
---
 configure | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
configure +3 -4
index 206d22c5153..002792d21dc 100755
@@ -7832,19 +7832,19 @@ case "$target_name" in
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips|mipsel)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips
     echo "TARGET_ABI_MIPSO32=y" >> $config_target_mak
   ;;
   mipsn32|mipsn32el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN32=y" >> $config_target_mak
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips64|mips64el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN64=y" >> $config_target_mak
--

2.20.1

#4 lore raw

пет, 20. мар 2020. у 18:08 Aleksandar Markovic <
[email protected]> је написао/ла:

Quoted 2 lines

пет, 20. мар 2020. у 12:45 Alex Bennée <[email protected]> је

написао/ла:

Quoted 9 lines
Quoted 9 lines

While debugging check-acceptance failures I found an instability in the mips64el test case. Briefly the test case:

  retry.py -n 100 -c -- ./mips64el-softmmu/qemu-system-mips64el \
    -display none -vga none -serial mon:stdio \
    -machine malta -kernel ./vmlinux-4.7.0-rc1.I6400 \
    -cpu I6400 -smp 8 -vga std \
    -append "printk.time=0 clocksource=GIC console=tty0 console=ttyS0

panic=-1" \

Quoted 7 lines
Quoted 2 lines
    --no-reboot

Thank for the findings!

Could you perhaps attach or link to "retry.py"?

Is this the script you used:

https://github.com/stsquad/retry/blob/master/retry.py

Quoted 18 lines

Did you run this particular test for the first time now, or it used to pass before?

Thanks,
Aleksandar

Quoted 12 lines

Reports about a 9% failure rate:

Results summary:
0: 91 times (91.00%), avg time 5.547 (0.45 varience/0.67 deviation) -6: 9 times (9.00%), avg time 3.394 (0.02 varience/0.13 deviation) Ran command 100 times, 91 passes

When re-run with "--accel tcg,thread=single" the instability goes away.

Results summary:
0: 100 times (100.00%), avg time 17.318 (249.76 varience/15.80

deviation)

Quoted 50 lines
Quoted 50 lines

Ran command 100 times, 100 passes

Which seems to indicate there is some aspect of the MIPS MTTCG fixes that has been missed. Ideally we would fix that but I'm afraid I don't have time to investigate and am not super familiar with the architecture anyway.

I've disabled all the mips guests as I assume it's a fundamental synchronisation primitive that is broken but I haven't tested them all (there are a lot!).

Cc Aleksandar Markovic <[email protected]>
Cc Aurelien Jarno <[email protected]>
Cc Aleksandar Rikalo <[email protected]>
Cc Philippe Mathieu-Daudé <[email protected]>
---
 configure | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
configure +3 -4
index 206d22c5153..002792d21dc 100755
@@ -7832,19 +7832,19 @@ case "$target_name" in
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips|mipsel)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips
     echo "TARGET_ABI_MIPSO32=y" >> $config_target_mak
   ;;
   mipsn32|mipsn32el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN32=y" >> $config_target_mak
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips64|mips64el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN64=y" >> $config_target_mak
--

2.20.1

#5 lore raw
In reply to Aleksandar Markovic
Quoted 13 lines · Aleksandar Markovic

пет, 20. мар 2020. у 12:45 Alex Bennée <[email protected]> је написао/ла:

Quoted 11 lines

While debugging check-acceptance failures I found an instability in the mips64el test case. Briefly the test case:

  retry.py -n 100 -c -- ./mips64el-softmmu/qemu-system-mips64el \
    -display none -vga none -serial mon:stdio \
    -machine malta -kernel ./vmlinux-4.7.0-rc1.I6400 \
    -cpu I6400 -smp 8 -vga std \
    -append "printk.time=0 clocksource=GIC console=tty0 console=ttyS0 panic=-1" \
    --no-reboot
Quoted 3 lines

Thank for the findings!

Could you perhaps attach or link to "retry.py"?

Sure - it's just a noddy python script which I use for repeated testing:

https://github.com/stsquad/retry

Quoted 2 lines

Did you run this particular test for the first time now, or it used to pass before?

I only noticed it since it was added to check-acceptance and has been flakey since added I think.

Quoted 66 lines

Thanks,
Aleksandar

Quoted 62 lines

Reports about a 9% failure rate:

Results summary:
0: 91 times (91.00%), avg time 5.547 (0.45 varience/0.67 deviation) -6: 9 times (9.00%), avg time 3.394 (0.02 varience/0.13 deviation) Ran command 100 times, 91 passes

When re-run with "--accel tcg,thread=single" the instability goes away.

Results summary:
0: 100 times (100.00%), avg time 17.318 (249.76 varience/15.80 deviation) Ran command 100 times, 100 passes

Which seems to indicate there is some aspect of the MIPS MTTCG fixes that has been missed. Ideally we would fix that but I'm afraid I don't have time to investigate and am not super familiar with the architecture anyway.

I've disabled all the mips guests as I assume it's a fundamental synchronisation primitive that is broken but I haven't tested them all (there are a lot!).

Cc Aleksandar Markovic <[email protected]>
Cc Aurelien Jarno <[email protected]>
Cc Aleksandar Rikalo <[email protected]>
Cc Philippe Mathieu-Daudé <[email protected]>
---
 configure | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)
configure +3 -4
index 206d22c5153..002792d21dc 100755
@@ -7832,19 +7832,19 @@ case "$target_name" in
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips|mipsel)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips
     echo "TARGET_ABI_MIPSO32=y" >> $config_target_mak
   ;;
   mipsn32|mipsn32el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN32=y" >> $config_target_mak
     echo "TARGET_ABI32=y" >> $config_target_mak
   ;;
   mips64|mips64el)
-    mttcg="yes"
+    mttcg="no"
     TARGET_ARCH=mips64
     TARGET_BASE_ARCH=mips
     echo "TARGET_ABI_MIPSN64=y" >> $config_target_mak
--

2.20.1

Signature · 1 line
Alex Bennée