[Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Philippe Mathieu-Daudé]

Xtensa cpu supported:
- dc232b
- dc233c
- csp

Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
---
based on http://lists.nongnu.org/archive/html/qemu-devel/2018-06/msg06782.html

Max: Do we need to use the overlay configuration?

This image allow the build of your dc232b tests (using [1]):

$ make check-tcg
...
  CC      vl.o
  CC      qmp.o
  LINK    xtensa-softmmu/qemu-system-xtensa
  BUILD   debian-xtensa-cross
  BUILD   TCG tests for xtensa-softmmu
  CHECK   debian-xtensa-cross
  BUILD   xtensa guest-tests with docker qemu:debian-xtensa-cross
  RUN     TCG tests for xtensa-softmmu
  CHECK   debian-xtensa-cross
  BUILD   xtensa guest-tests with docker qemu:debian-xtensa-cross
$ file xtensa-softmmu/tests/*tst
xtensa-softmmu/tests/test_bi.tst:        ELF 32-bit LSB executable, Tensilica Xtensa, version 1 (SYSV), statically linked, not stripped
xtensa-softmmu/tests/test_break.tst:     ELF 32-bit LSB executable, Tensilica Xtensa, version 1 (SYSV), statically linked, not stripped
...
xtensa-softmmu/tests/test_timer.tst:     ELF 32-bit LSB executable, Tensilica Xtensa, version 1 (SYSV), statically linked, not stripped
xtensa-softmmu/tests/test_windowed.tst:  ELF 32-bit LSB executable, Tensilica Xtensa, version 1 (SYSV), statically linked, not stripped

[1]: https://github.com/philmd/qemu/tree/testing/gcov-and-other-build-tweaks-v1c

 .../dockerfiles/debian-xtensa-cross.docker    | 31 +++++++++++++++++++
 1 file changed, 31 insertions(+)
 create mode 100644 tests/docker/dockerfiles/debian-xtensa-cross.docker

diff --git a/tests/docker/dockerfiles/debian-xtensa-cross.docker b/tests/docker/dockerfiles/debian-xtensa-cross.docker
new file mode 100644
index 0000000000..afd2ab9163
--- /dev/null
+++ b/tests/docker/dockerfiles/debian-xtensa-cross.docker
@@ -0,0 +1,31 @@
+#
+# Docker cross-compiler target
+#
+# This docker target builds on the debian stretch base image,
+# using a prebuilt toolchains for Xtensa cores from:
+# https://github.com/foss-xtensa/toolchain/releases
+#
+FROM debian:stretch-slim
+
+RUN apt-get update && \
+    DEBIAN_FRONTEND=noninteractive apt install -yy eatmydata
+RUN DEBIAN_FRONTEND=noninteractive eatmydata \
+    apt-get install -y --no-install-recommends \
+        bison \
+        build-essential \
+        ca-certificates \
+        curl \
+        flex \
+        gettext \
+        git \
+        python-minimal
+
+ENV CPU_LIST csp dc232b dc233c
+ENV TOOLCHAIN_RELEASE 2018.02
+
+RUN for cpu in $CPU_LIST; do \
+        curl -#SL http://github.com/foss-xtensa/toolchain/releases/download/$TOOLCHAIN_RELEASE/x86_64-$TOOLCHAIN_RELEASE-xtensa-$cpu-elf.tar.gz \
+        | tar -xzC /opt; \
+    done
+
+ENV PATH $PATH:/opt/$TOOLCHAIN_RELEASE/xtensa-dc232b-elf/bin:/opt/$TOOLCHAIN_RELEASE/xtensa-dc233c-elf/bin:/opt/$TOOLCHAIN_RELEASE/xtensa-csp-elf/bin
-- 
2.18.0

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Philippe Mathieu-Daudé]

On 06/23/2018 06:53 PM, Philippe Mathieu-Daudé wrote:
> Xtensa cpu supported:
> - dc232b
> - dc233c
> - csp
> 
> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
> ---
> based on http://lists.nongnu.org/archive/html/qemu-devel/2018-06/msg06782.html
> 
> Max: Do we need to use the overlay configuration?
> 
> This image allow the build of your dc232b tests (using [1]):
> 
> $ make check-tcg
> ...
>   CC      vl.o
>   CC      qmp.o
>   LINK    xtensa-softmmu/qemu-system-xtensa
>   BUILD   debian-xtensa-cross
>   BUILD   TCG tests for xtensa-softmmu
>   CHECK   debian-xtensa-cross
>   BUILD   xtensa guest-tests with docker qemu:debian-xtensa-cross
>   RUN     TCG tests for xtensa-softmmu
>   CHECK   debian-xtensa-cross
>   BUILD   xtensa guest-tests with docker qemu:debian-xtensa-cross
> $ file xtensa-softmmu/tests/*tst
> xtensa-softmmu/tests/test_bi.tst:        ELF 32-bit LSB executable, Tensilica Xtensa, version 1 (SYSV), statically linked, not stripped
> xtensa-softmmu/tests/test_break.tst:     ELF 32-bit LSB executable, Tensilica Xtensa, version 1 (SYSV), statically linked, not stripped
> ...
> xtensa-softmmu/tests/test_timer.tst:     ELF 32-bit LSB executable, Tensilica Xtensa, version 1 (SYSV), statically linked, not stripped
> xtensa-softmmu/tests/test_windowed.tst:  ELF 32-bit LSB executable, Tensilica Xtensa, version 1 (SYSV), statically linked, not stripped
> 
> [1]: https://github.com/philmd/qemu/tree/testing/gcov-and-other-build-tweaks-v1c

Max, some tests never end, is that normal?

$ make check-tcg
...
  CC      vl.o
  CC      qmp.o
  LINK    xtensa-softmmu/qemu-system-xtensa
  BUILD   debian-xtensa-cross
  BUILD   TCG tests for xtensa-softmmu
  CHECK   debian-xtensa-cross
  BUILD   xtensa guest-tests with docker qemu:debian-xtensa-cross
  RUN     TCG tests for xtensa-softmmu
  CHECK   debian-xtensa-cross
  BUILD   xtensa guest-tests with docker qemu:debian-xtensa-cross
  RUN     tests for xtensa
QEMU 2.12.50 monitor - type 'help' for more information
(qemu) info registers
PC=00007fff

        LBEG=d0001b1c         LEND=d0001b26       LCOUNT=00000000
  SAR=00000000
     LITBASE=00000000    SCOMPARE1=00000000        ACCLO=00000000
ACCHI=00000000
         MR0=00000000          MR1=00000000          MR2=00000000
  MR3=00000000
 WINDOW_BASE=00000000 WINDOW_START=00000001     PTEVADDR=80000000
 MMID=00000000
       RASID=04030201      ITLBCFG=00000000      DTLBCFG=00000000
IBREAKENABLE=00000000
      MEMCTL=00000000          DDR=00000000     IBREAKA0=00000000
IBREAKA1=00000000
    DBREAKA0=00000000     DBREAKA1=00000000     DBREAKC0=00000000
DBREAKC1=00000000
   CONFIGID0=c56307fe         EPC1=000019f9         EPC2=00000000
 EPC3=00000000
        EPC4=00000000         EPC5=00000000         EPC6=00000000
 EPC7=00000000
        DEPC=d000128a         EPS2=00000000         EPS3=00000000
 EPS4=00000000
        EPS5=00000000         EPS6=00000000         EPS7=00000000
CONFIGID1=0d40beef
    EXCSAVE1=0004004f     EXCSAVE2=00000000     EXCSAVE3=00000000
EXCSAVE4=00000000
    EXCSAVE5=00000000     EXCSAVE6=00000000     EXCSAVE7=00000000
CPENABLE=00000000
      INTSET=00000000     INTCLEAR=00000000    INTENABLE=00000000
   PS=0004000f
     VECBASE=d0000000     EXCCAUSE=00000018   DEBUGCAUSE=00000000
CCOUNT=00000000
        PRID=00000000       ICOUNT=00000000  ICOUNTLEVEL=00000000
EXCVADDR=00400001
   CCOMPARE0=00000000    CCOMPARE1=00000000    CCOMPARE2=00000000
MISC0=00000000
       MISC1=00000000        MISC2=00000000        MISC3=00000000

EXPSTATE=00000000 THREADPTR=00000000

 A00=d000058d  A01=00000018  A02=00007fff  A03=00008000
 A04=00000006  A05=00000000  A06=00000000  A07=00000000
 A08=00000000  A09=00000000  A10=00000000  A11=00000000
 A12=00000000  A13=00000000  A14=00000000  A15=00000000

AR00=d000058d AR01=00000018 AR02=00007fff AR03=00008000 <=
AR04=00000006 AR05=00000000 AR06=00000000 AR07=00000000
AR08=00000000 AR09=00000000 AR10=00000000 AR11=00000000
AR12=00000000 AR13=00000000 AR14=00000000 AR15=00000000
AR16=00000000 AR17=00000000 AR18=00000000 AR19=00000000
AR20=00000000 AR21=00000000 AR22=00000000 AR23=00000000
AR24=00000000 AR25=00000000 AR26=00000000 AR27=00000000
AR28=00000000 AR29=00000000 AR30=00000000 AR31=00000000
# on another terminal: killall -9 qemu-system-xtensa
make[2]: *** [/source/qemu/tests/tcg/xtensa/Makefile.system:76:
run-test_mmu.tst] Killed
make[1]: *** [/source/qemu/tests/tcg/Makefile.include:76:
run-guest-tests] Error 2
make: *** [/source/qemu/tests/Makefile.include:946:
run-tcg-tests-xtensa-softmmu] Error 2

This one works:

xtensa-softmmu/tests$ ../qemu-system-xtensa -M sim -cpu dc232b
-nographic -semihosting -icount 6 -kernel test_mul32.tst -d in_asm
QEMU 2.12.50 monitor - type 'help' for more information
(qemu) ----------------
IN:
0xfe000000:  j	0xfe000008

----------------
IN:
0xfe000008:  l32r	a2, 0xfe000004
0xfe00000b:  jx	a2

----------------
IN:
0xd0000404:  movi.n	a2, 1
0xd0000406:  wsr.windowstart	a2

----------------
IN:
0xd0000409:  movi.n	a2, 0
0xd000040b:  wsr.windowbase	a2

----------------
IN:
0xd000040e:  l32r	a1, 0xd00003e8
0xd0000411:  l32r	a2, 0xd00003ec
0xd0000414:  wsr.ps	a2

----------------
IN:
0xd0000417:  isync
0xd000041a:  call0	0xd0000424

----------------
IN:
0xd0000424:  l32r	a2, 0xd00003f0
0xd0000427:  mov.n	a3, a2
0xd0000429:  l32r	a4, 0xd00003f4
0xd000042c:  l32r	a6, 0xd00003f8
0xd000042f:  mull	a5, a2, a4
0xd0000432:  beq	a5, a6, 0xd0000442

----------------
IN:
0xd0000442:  nop.n
0xd0000444:  mull	a2, a2, a4
0xd0000447:  beq	a2, a6, 0xd0000458

----------------
IN:
0xd0000458:  nop.n
0xd000045a:  mull	a3, a4, a3
0xd000045d:  beq	a3, a6, 0xd000046d

----------------
IN:
0xd000046d:  nop.n
0xd000046f:  l32r	a2, 0xd00003ec
0xd0000472:  wsr.ps	a2

----------------
IN:
0xd0000475:  isync
0xd0000478:  l32r	a2, 0xd00003fc
0xd000047b:  l32i.n	a3, a2, 0
0xd000047d:  addi.n	a3, a3, 1
0xd000047f:  s32i.n	a3, a2, 0
0xd0000481:  l32r	a2, 0xd00003ec
0xd0000484:  wsr.ps	a2

----------------
IN:
0xd0000487:  isync
0xd000048a:  l32r	a0, 0xd00003fc
0xd000048d:  l32i	a2, a0, 0
0xd0000490:  l32r	a0, 0xd0000400
0xd0000493:  sub	a2, a2, a0
0xd0000496:  movi	a3, 0
0xd0000499:  loopnez	a2, 0xd00004a4

----------------
IN:
0xd000049c:  l8ui	a2, a0, 0
0xd000049f:  or	a3, a3, a2
0xd00004a2:  addi.n	a0, a0, 1

----------------
IN:
0xd00004a4:  movi.n	a2, 1
0xd00004a6:  simcall
0xd00004a9:  ill

exit(0) simcall


But this one for example never exit:

xtensa-softmmu/tests$ ../qemu-system-xtensa -M sim -cpu dc232b
-nographic -semihosting -icount 6 -kernel ./test_mmu.tst -d in_asm
...
----------------
IN:
0xd0001b0a:  wdtlb	a2, a3

----------------
IN:
0xd0001b0d:  l32r	a2, 0xd0000540
0xd0001b10:  l32r	a3, 0xd0000544
0xd0001b13:  l32r	a4, 0xd0000534
0xd0001b16:  sub	a4, a4, a3
0xd0001b19:  loop	a4, 0xd0001b26

----------------
IN:
0xd0001b1c:  l8ui	a5, a3, 0
0xd0001b1f:  s8i	a5, a2, 0
0xd0001b22:  addi.n	a2, a2, 1
0xd0001b24:  addi.n	a3, a3, 1

----------------
IN:
0xd0001b26:  l32r	a2, 0xd0000540
0xd0001b29:  l32r	a3, 0xd000053c
0xd0001b2c:  jx	a2

----------------
IN:
0x00007fff:  l32i	a2, a3, 0

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Philippe Mathieu-Daudé]

> But this one for example never exit:
> 
> xtensa-softmmu/tests$ ../qemu-system-xtensa -M sim -cpu dc232b
> -nographic -semihosting -icount 6 -kernel ./test_mmu.tst -d in_asm
> ...
> ----------------
> IN:
> 0xd0001b0a:  wdtlb	a2, a3
> 
> ----------------
> IN:
> 0xd0001b0d:  l32r	a2, 0xd0000540
> 0xd0001b10:  l32r	a3, 0xd0000544
> 0xd0001b13:  l32r	a4, 0xd0000534
> 0xd0001b16:  sub	a4, a4, a3
> 0xd0001b19:  loop	a4, 0xd0001b26
> 
> ----------------
> IN:
> 0xd0001b1c:  l8ui	a5, a3, 0
> 0xd0001b1f:  s8i	a5, a2, 0
> 0xd0001b22:  addi.n	a2, a2, 1
> 0xd0001b24:  addi.n	a3, a3, 1
> 
> ----------------
> IN:
> 0xd0001b26:  l32r	a2, 0xd0000540
> 0xd0001b29:  l32r	a3, 0xd000053c
> 0xd0001b2c:  jx	a2
> 
> ----------------
> IN:
> 0x00007fff:  l32i	a2, a3, 0
> 

Also it never quit (return to shell):

xtensa-softmmu/tests$ ../qemu-system-xtensa -M sim -cpu dc232b
-nographic -semihosting -icount 6 -kernel ./test_mmu.tst
QEMU 2.12.50 monitor - type 'help' for more information
(qemu) q
HANG

(gdb) info threads
  Id   Target Id         Frame
* 1    Thread 0x7fdfead6bf80 (LWP 1645) 0x00007fdfce6dc4ec in
futex_wait_cancelable (private=<optimized out>, expected=0,
    futex_word=0x55b9940ae288 <qemu_pause_cond+40>) at
../sysdeps/unix/sysv/linux/futex-internal.h:88
  2    Thread 0x7fdfbfc79700 (LWP 1646) syscall () at
../sysdeps/unix/sysv/linux/x86_64/syscall.S:38
  3    Thread 0x7fdfbf478700 (LWP 1647) 0x00007fdfce4016d6 in __GI_ppoll
(fds=0x7fdfb0000b20, nfds=1, timeout=<optimized out>,
    timeout@entry=0x0, sigmask=sigmask@entry=0x0) at
../sysdeps/unix/sysv/linux/ppoll.c:39
  4    Thread 0x7fdfbcc77700 (LWP 1648) qemu_spin_lock (spin=<optimized
out>) at /source/qemu/include/qemu/thread.h:111

(gdb) thread apply all bt

Thread 4 (Thread 0x7fdfbcc77700 (LWP 1648)):
#0  0x000055b993a62fe0 in qemu_spin_lock (spin=<optimized out>) at
/source/qemu/include/qemu/thread.h:111
#1  0x000055b993a62fe0 in page_lock (pd=<optimized out>) at
/source/qemu/accel/tcg/translate-all.c:655
#2  0x000055b993a62fe0 in page_lock_pair
(ret_p1=ret_p1@entry=0x7fdfbcc74550, phys1=phys1@entry=100667391,
ret_p2=ret_p2@entry=0x7fdfbcc74558, phys2=phys2@entry=100663296,
alloc=alloc@entry=1) at /source/qemu/accel/tcg/translate-all.c:873
#3  0x000055b993a63643 in tb_link_page (phys_page2=100663296,
phys_pc=100667391, tb=<optimized out>)
    at /source/qemu/accel/tcg/translate-all.c:1602
#4  0x000055b993a63643 in tb_gen_code (cpu=cpu@entry=0x55b9943eb760,
pc=pc@entry=32767, cs_base=cs_base@entry=0, flags=flags@entry=98304,
cflags=cflags@entry=131072) at /source/qemu/accel/tcg/translate-all.c:1794
#5  0x000055b993a61937 in tb_find (cf_mask=131072, tb_exit=0,
last_tb=0x0, cpu=0x18000) at /source/qemu/accel/tcg/cpu-exec.c:401
#6  0x000055b993a61937 in cpu_exec (cpu=cpu@entry=0x55b9943eb760) at
/source/qemu/accel/tcg/cpu-exec.c:721
#7  0x000055b993a307a1 in tcg_cpu_exec (cpu=0x55b9943eb760) at
/source/qemu/cpus.c:1362
#8  0x000055b993a307a1 in qemu_tcg_rr_cpu_thread_fn (arg=<optimized
out>) at /source/qemu/cpus.c:1461
#9  0x00007fdfce6d65aa in start_thread (arg=0x7fdfbcc77700) at
pthread_create.c:463
#10 0x00007fdfce40bcbf in clone () at
../sysdeps/unix/sysv/linux/x86_64/clone.S:95

Thread 3 (Thread 0x7fdfbf478700 (LWP 1647)):
#0  0x00007fdfce4016d6 in __GI_ppoll (fds=0x7fdfb0000b20, nfds=1,
timeout=<optimized out>, timeout@entry=0x0, sigmask=sigmask@entry=0x0)
    at ../sysdeps/unix/sysv/linux/ppoll.c:39
#1  0x000055b993c33109 in ppoll (__ss=0x0, __timeout=0x0,
__nfds=<optimized out>, __fds=<optimized out>)
    at /usr/include/x86_64-linux-gnu/bits/poll2.h:77
#2  0x000055b993c33109 in qemu_poll_ns (fds=<optimized out>,
nfds=<optimized out>, timeout=<optimized out>)
    at /source/qemu/util/qemu-timer.c:322
#3  0x000055b993c34dce in aio_poll (ctx=0x55b9943b35c0,
blocking=blocking@entry=true) at /source/qemu/util/aio-posix.c:629
#4  0x000055b993a9897e in iothread_run (opaque=0x55b9943b3360) at
/source/qemu/iothread.c:64
#5  0x00007fdfce6d65aa in start_thread (arg=0x7fdfbf478700) at
pthread_create.c:463
#6  0x00007fdfce40bcbf in clone () at
../sysdeps/unix/sysv/linux/x86_64/clone.S:95

Thread 2 (Thread 0x7fdfbfc79700 (LWP 1646)):
#0  0x00007fdfce4068f9 in syscall () at
../sysdeps/unix/sysv/linux/x86_64/syscall.S:38
#1  0x000055b993c370eb in qemu_futex_wait (val=<optimized out>,
f=<optimized out>) at /source/qemu/include/qemu/futex.h:29
#2  0x000055b993c370eb in qemu_event_wait (ev=ev@entry=0x55b9940e49c8
<rcu_call_ready_event>)
    at /source/qemu/util/qemu-thread-posix.c:445
#3  0x000055b993c476a8 in call_rcu_thread (opaque=<optimized out>) at
/source/qemu/util/rcu.c:261
#4  0x00007fdfce6d65aa in start_thread (arg=0x7fdfbfc79700) at
pthread_create.c:463
#5  0x00007fdfce40bcbf in clone () at
../sysdeps/unix/sysv/linux/x86_64/clone.S:95

Thread 1 (Thread 0x7fdfead6bf80 (LWP 1645)):
#0  0x00007fdfce6dc4ec in futex_wait_cancelable (private=<optimized
out>, expected=0, futex_word=0x55b9940ae288 <qemu_pause_cond+40>)
    at ../sysdeps/unix/sysv/linux/futex-internal.h:88
#1  0x00007fdfce6dc4ec in __pthread_cond_wait_common (abstime=0x0,
mutex=0x55b9940ae2e0 <qemu_global_mutex>, cond=0x55b9940ae260
<qemu_pause_cond>) at pthread_cond_wait.c:502
#2  0x00007fdfce6dc4ec in __pthread_cond_wait
(cond=cond@entry=0x55b9940ae260 <qemu_pause_cond>,
mutex=mutex@entry=0x55b9940ae2e0 <qemu_global_mutex>) at
pthread_cond_wait.c:655
#3  0x000055b993c36c9f in qemu_cond_wait_impl
(cond=cond@entry=0x55b9940ae260 <qemu_pause_cond>,
mutex=mutex@entry=0x55b9940ae2e0 <qemu_global_mutex>,
file=file@entry=0x55b993c52bdb "/source/qemu/cpus.c", line=line@entry=1809)
    at /source/qemu/util/qemu-thread-posix.c:164
#4  0x000055b993a30ce0 in pause_all_vcpus () at /source/qemu/cpus.c:1809
#5  0x000055b993a30d6a in do_vm_stop (state=RUN_STATE_SHUTDOWN,
send_stop=<optimized out>) at /source/qemu/cpus.c:1010
#6  0x000055b9939ef579 in main (argc=<optimized out>, argv=<optimized
out>, envp=<optimized out>) at /source/qemu/vl.c:4605

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Max Filippov]

On Sat, Jun 23, 2018 at 2:57 PM, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
> On 06/23/2018 06:53 PM, Philippe Mathieu-Daudé wrote:
> Max, some tests never end, is that normal?

No, you've just found and reported a bug. Thank you.
Bisection points to the following commit:

commit 0b5c91f74f3c83a36f37740969df8c775c997e69
Author: Emilio G. Cota <cota@braap.org>
Date:   Wed Jul 26 20:22:51 2017 -0400

    translate-all: use per-page locking in !user-mode

Bisection log:

git bisect start
# bad: [f28d0dfdce5754d80d2a5993fff2f5312b32cac1] tcg: fix
--disable-tcg build breakage
git bisect bad f28d0dfdce5754d80d2a5993fff2f5312b32cac1
# good: [c74e62ee3e2dc2955e07d004c71badecb68a84eb] Merge
remote-tracking branch 'remotes/rth/tags/cota-target-pull-request'
into staging
git bisect good c74e62ee3e2dc2955e07d004c71badecb68a84eb
# good: [f67c9b693ae3d03305162623a043ba4067a3c00c] Merge
remote-tracking branch 'remotes/mst/tags/for_upstream' into staging
git bisect good f67c9b693ae3d03305162623a043ba4067a3c00c
# good: [21f402093c5ee7363f5ba56916cd5c651b424fef] iommu: Add IOMMU
index concept to IOMMU API
git bisect good 21f402093c5ee7363f5ba56916cd5c651b424fef
# good: [7a5d936b6fc0cf262db86af0df7999ac904f8495] docker: docker.py
wrap StringIO import for python3
git bisect good 7a5d936b6fc0cf262db86af0df7999ac904f8495
# bad: [de44c044420d1139480fa50c2d5be19223391218] Merge
remote-tracking branch
'remotes/stsquad/tags/pull-tcg-testing-revivial-210618-2' into staging
git bisect bad de44c044420d1139480fa50c2d5be19223391218
# good: [7e97017e7db3c0ce552cf1a58df8bb97bde29c98] tests/tcg/Makefile:
update to be called from Makefile.target
git bisect good 7e97017e7db3c0ce552cf1a58df8bb97bde29c98
# bad: [33836a731562e3d07b3a83f26e81c6b1482d216c] Merge
remote-tracking branch 'remotes/rth/tags/pull-tcg-20180615' into
staging
git bisect bad 33836a731562e3d07b3a83f26e81c6b1482d216c
# good: [45c73de594414904b0d6a7ade70fb4514d35f79c] translate-all: move
tb_invalidate_phys_page_range up in the file
git bisect good 45c73de594414904b0d6a7ade70fb4514d35f79c
# bad: [194125e3ebd553acb02aaf3797a4f0387493fe94] translate-all:
protect TB jumps with a per-destination-TB lock
git bisect bad 194125e3ebd553acb02aaf3797a4f0387493fe94
# bad: [6d9abf85d538731ccff25fc29d7fa938115b1a80] translate-all: add
page_locked assertions
git bisect bad 6d9abf85d538731ccff25fc29d7fa938115b1a80
# bad: [0b5c91f74f3c83a36f37740969df8c775c997e69] translate-all: use
per-page locking in !user-mode
git bisect bad 0b5c91f74f3c83a36f37740969df8c775c997e69
# first bad commit: [0b5c91f74f3c83a36f37740969df8c775c997e69]
translate-all: use per-page locking in !user-mode

Emilio, could you please take a look? The following test locks up QEMU:

qemu-system-xtensa -M sim -cpu dc232b -nographic -semihosting -icount
6  -kernel ./test_mmu.tst

test_mmu.tst binary may be found here:
http://jcmvbkbc.spb.ru/~dumb/tmp/201806232022/test_mmu.tst

-- 
Thanks.
-- Max

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Emilio G. Cota]

On Sat, Jun 23, 2018 at 20:24:09 -0700, Max Filippov wrote:
> # first bad commit: [0b5c91f74f3c83a36f37740969df8c775c997e69]
> translate-all: use per-page locking in !user-mode
> 
> Emilio, could you please take a look? The following test locks up QEMU:
> 
> qemu-system-xtensa -M sim -cpu dc232b -nographic -semihosting -icount
> 6  -kernel ./test_mmu.tst
> 
> test_mmu.tst binary may be found here:
> http://jcmvbkbc.spb.ru/~dumb/tmp/201806232022/test_mmu.tst

Thanks for the test case.

I submitted a patch that fixes this for me--see this thread:
  https://lists.gnu.org/archive/html/qemu-devel/2018-06/msg07236.html

		Emilio

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Philippe Mathieu-Daudé]

On 06/25/2018 01:59 PM, Emilio G. Cota wrote:
> On Sat, Jun 23, 2018 at 20:24:09 -0700, Max Filippov wrote:
>> # first bad commit: [0b5c91f74f3c83a36f37740969df8c775c997e69]
>> translate-all: use per-page locking in !user-mode
>>
>> Emilio, could you please take a look? The following test locks up QEMU:
>>
>> qemu-system-xtensa -M sim -cpu dc232b -nographic -semihosting -icount
>> 6  -kernel ./test_mmu.tst
>>
>> test_mmu.tst binary may be found here:
>> http://jcmvbkbc.spb.ru/~dumb/tmp/201806232022/test_mmu.tst
> 
> Thanks for the test case.
> 
> I submitted a patch that fixes this for me--see this thread:
>   https://lists.gnu.org/archive/html/qemu-devel/2018-06/msg07236.html

Thanks, this works :)

Alex nicer test output using:

tests/tcg/xtensa/Makefile.system
 run-%.tst: %.tst
-       $(SIM) $(SIMFLAGS) ./$<
+       $(SIM) -monitor null $(SIMFLAGS) ./$<

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Max Filippov]

On Mon, Jun 25, 2018 at 10:10 AM, Philippe Mathieu-Daudé
<f4bug@amsat.org> wrote:
> On 06/25/2018 01:59 PM, Emilio G. Cota wrote:
>> On Sat, Jun 23, 2018 at 20:24:09 -0700, Max Filippov wrote:
>>> # first bad commit: [0b5c91f74f3c83a36f37740969df8c775c997e69]
>>> translate-all: use per-page locking in !user-mode
>>>
>>> Emilio, could you please take a look? The following test locks up QEMU:
>>>
>>> qemu-system-xtensa -M sim -cpu dc232b -nographic -semihosting -icount
>>> 6  -kernel ./test_mmu.tst
>>>
>>> test_mmu.tst binary may be found here:
>>> http://jcmvbkbc.spb.ru/~dumb/tmp/201806232022/test_mmu.tst
>>
>> Thanks for the test case.
>>
>> I submitted a patch that fixes this for me--see this thread:
>>   https://lists.gnu.org/archive/html/qemu-devel/2018-06/msg07236.html
>
> Thanks, this works :)
>
> Alex nicer test output using:
>
> tests/tcg/xtensa/Makefile.system
>  run-%.tst: %.tst
> -       $(SIM) $(SIMFLAGS) ./$<
> +       $(SIM) -monitor null $(SIMFLAGS) ./$<

This makefile is also usable with another simulator (xtensa ISS) which would
not understand this option. Could it be added to the proper SIMFLAGS instead?

-- 
Thanks.
-- Max

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Philippe Mathieu-Daudé]

On 06/25/2018 02:18 PM, Max Filippov wrote:
> On Mon, Jun 25, 2018 at 10:10 AM, Philippe Mathieu-Daudé
> <f4bug@amsat.org> wrote:
>> On 06/25/2018 01:59 PM, Emilio G. Cota wrote:
>>> On Sat, Jun 23, 2018 at 20:24:09 -0700, Max Filippov wrote:
>>>> # first bad commit: [0b5c91f74f3c83a36f37740969df8c775c997e69]
>>>> translate-all: use per-page locking in !user-mode
>>>>
>>>> Emilio, could you please take a look? The following test locks up QEMU:
>>>>
>>>> qemu-system-xtensa -M sim -cpu dc232b -nographic -semihosting -icount
>>>> 6  -kernel ./test_mmu.tst
>>>>
>>>> test_mmu.tst binary may be found here:
>>>> http://jcmvbkbc.spb.ru/~dumb/tmp/201806232022/test_mmu.tst
>>>
>>> Thanks for the test case.
>>>
>>> I submitted a patch that fixes this for me--see this thread:
>>>   https://lists.gnu.org/archive/html/qemu-devel/2018-06/msg07236.html
>>
>> Thanks, this works :)
>>
>> Alex nicer test output using:
>>
>> tests/tcg/xtensa/Makefile.system
>>  run-%.tst: %.tst
>> -       $(SIM) $(SIMFLAGS) ./$<
>> +       $(SIM) -monitor null $(SIMFLAGS) ./$<
> 
> This makefile is also usable with another simulator (xtensa ISS) which would
> not understand this option. Could it be added to the proper SIMFLAGS instead?

OK, good to know.

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Max Filippov]

On Sat, Jun 23, 2018 at 2:53 PM, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
> Xtensa cpu supported:
> - dc232b
> - dc233c
> - csp
>
> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
> ---
> based on http://lists.nongnu.org/archive/html/qemu-devel/2018-06/msg06782.html
>
> Max: Do we need to use the overlay configuration?

You'd need to use overlay if you'd build the toolchain.
Prebuilt toolchains are already configured for specific processors.
I didn't try to run it in docker, but it looks like you're doing the
right thing,
especially that you see some tests passing.

Reviewed-by: Max Filippov <jcmvbkbc@gmail.com>
-- 
Thanks.
-- Max

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Philippe Mathieu-Daudé]

On 06/24/2018 12:28 AM, Max Filippov wrote:
> On Sat, Jun 23, 2018 at 2:53 PM, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
>> Xtensa cpu supported:
>> - dc232b
>> - dc233c
>> - csp
>>
>> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>> ---
>> based on http://lists.nongnu.org/archive/html/qemu-devel/2018-06/msg06782.html
>>
>> Max: Do we need to use the overlay configuration?
> 
> You'd need to use overlay if you'd build the toolchain.
> Prebuilt toolchains are already configured for specific processors.

I now understand.

These 3 cpu toolchains are now handily accesible:

$ docker run --rm -it qemu:debian-xtensa-cross
root@e0a99f962594:/# xtensa-<tab><tab>
xtensa-csp-elf-addr2line        xtensa-dc232b-elf-gcov-dump
xtensa-csp-elf-ar               xtensa-dc232b-elf-gcov-tool
xtensa-csp-elf-as               xtensa-dc232b-elf-gprof
xtensa-csp-elf-c++filt          xtensa-dc232b-elf-ld
xtensa-csp-elf-cc               xtensa-dc232b-elf-ld.bfd
xtensa-csp-elf-cpp              xtensa-dc232b-elf-nm
xtensa-csp-elf-ct-ng.config     xtensa-dc232b-elf-objcopy
xtensa-csp-elf-elfedit          xtensa-dc232b-elf-objdump
xtensa-csp-elf-gcc              xtensa-dc232b-elf-ranlib
xtensa-csp-elf-gcc-7.3.0        xtensa-dc232b-elf-readelf
xtensa-csp-elf-gcc-ar           xtensa-dc232b-elf-size
xtensa-csp-elf-gcc-nm           xtensa-dc232b-elf-strings
xtensa-csp-elf-gcc-ranlib       xtensa-dc232b-elf-strip
xtensa-csp-elf-gcov             xtensa-dc233c-elf-addr2line
xtensa-csp-elf-gcov-dump        xtensa-dc233c-elf-ar
xtensa-csp-elf-gcov-tool        xtensa-dc233c-elf-as
xtensa-csp-elf-gprof            xtensa-dc233c-elf-c++filt
xtensa-csp-elf-ld               xtensa-dc233c-elf-cc
xtensa-csp-elf-ld.bfd           xtensa-dc233c-elf-cpp
xtensa-csp-elf-nm               xtensa-dc233c-elf-ct-ng.config
xtensa-csp-elf-objcopy          xtensa-dc233c-elf-elfedit
xtensa-csp-elf-objdump          xtensa-dc233c-elf-gcc
xtensa-csp-elf-ranlib           xtensa-dc233c-elf-gcc-7.3.0
xtensa-csp-elf-readelf          xtensa-dc233c-elf-gcc-ar
xtensa-csp-elf-size             xtensa-dc233c-elf-gcc-nm
xtensa-csp-elf-strings          xtensa-dc233c-elf-gcc-ranlib
xtensa-csp-elf-strip            xtensa-dc233c-elf-gcov
xtensa-dc232b-elf-addr2line     xtensa-dc233c-elf-gcov-dump
xtensa-dc232b-elf-ar            xtensa-dc233c-elf-gcov-tool
xtensa-dc232b-elf-as            xtensa-dc233c-elf-gprof
xtensa-dc232b-elf-c++filt       xtensa-dc233c-elf-ld
xtensa-dc232b-elf-cc            xtensa-dc233c-elf-ld.bfd
xtensa-dc232b-elf-cpp           xtensa-dc233c-elf-nm
xtensa-dc232b-elf-ct-ng.config  xtensa-dc233c-elf-objcopy
xtensa-dc232b-elf-elfedit       xtensa-dc233c-elf-objdump
xtensa-dc232b-elf-gcc           xtensa-dc233c-elf-ranlib
xtensa-dc232b-elf-gcc-7.3.0     xtensa-dc233c-elf-readelf
xtensa-dc232b-elf-gcc-ar        xtensa-dc233c-elf-size
xtensa-dc232b-elf-gcc-nm        xtensa-dc233c-elf-strings
xtensa-dc232b-elf-gcc-ranlib    xtensa-dc233c-elf-strip
xtensa-dc232b-elf-gcov
root@e0a99f962594:/# xtensa-

As you said, dc232b is well covered in QEMU and has tests.

I appreciate including the csp cpu because it is big-endian.

dc233c is less useful for testing (closer to the dc232b).

> I didn't try to run it in docker, but it looks like you're doing the
> right thing,
> especially that you see some tests passing.
> 
> Reviewed-by: Max Filippov <jcmvbkbc@gmail.com>

Thanks!

Phil.

Re: [Qemu-devel] [RFC PATCH] docker: Add debian-xtensa-cross image

[Max Filippov]

On Sat, Jun 23, 2018 at 9:04 PM, Philippe Mathieu-Daudé <f4bug@amsat.org> wrote:
> I appreciate including the csp cpu because it is big-endian.

csp is little endian. FSF was big endian, but the name no longer refers
to a specific processor, so it is also not very useful. Perhaps I just
need to add a well-defined big endian xtensa configuration to QEMU
and publish the overlay and prebuilt toolchain for it.

-- 
Thanks.
-- Max