Re: [Qemu-devel] [PATCH v2 05/11] translate-all: exit cpu_restore_state early if translating

[Alex Bennée]

Richard Henderson <address@hidden> writes:

> On 03/03/2017 06:53 AM, Alex Bennée wrote:
>> The translation code uses cpu_ld*_code which can trigger a tlb_fill
>> which if it fails will attempt a fault resolution. This never works
>> during translation as the TB being generated hasn't been added yet.
>> However with the new locking regime we end up double locking the
>> tb_lock(). As the tcg_ctx.cpu is only set during translation we use
>> this to short circuit the restore code and return with a fail.
>
> We *should* have retaddr == 0 for this case, which indicates that we
> should not attempt to restore state.  Are you seeing a non-zero value?

Actually looking at xtensa I see:

  Attempt to resolve CPU state @ 0x0 while translating

So maybe I should check just that - but I don't see where we ensure we
always pass zero.

>
> Hmm.. Or rather we should not have called cpu_restore_state in the
> first place. What is the call chain leading to this point?


Thread 3 "qemu-system-xte" hit Breakpoint 2, cpu_restore_state (address@hidden, 
address@hidden) at /home/alex/lsrc/qemu/qemu.git/translate-all.c:338
338             qemu_log_mask(LOG_UNIMP, "Attempt to resolve CPU state @ 0x%" 
PRIxPTR
#0  0x00005555555e3712 in cpu_restore_state (address@hidden, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/translate-all.c:338
#1  0x000055555564cb38 in tlb_fill (address@hidden, address@hidden, 
address@hidden, address@hidden, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/target/xtensa/op_helper.c:73
#2  0x000055555562d604 in helper_ret_ldb_cmmu (address@hidden, addr=537034752, 
oi=<optimised out>, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/softmmu_template.h:127
#3  0x00005555556448df in gen_intermediate_code (retaddr=0, ptr=<optimised 
out>, env=0x55555603a890) at 
/home/alex/lsrc/qemu/qemu.git/include/exec/cpu_ldst_template.h:102
#4  0x00005555556448df in gen_intermediate_code (ptr=<optimised out>, 
env=0x55555603a890) at 
/home/alex/lsrc/qemu/qemu.git/include/exec/cpu_ldst_template.h:114
#5  0x00005555556448df in gen_intermediate_code (dc=0x7fffcca0f4d0, 
env=0x55555603a890) at 
/home/alex/lsrc/qemu/qemu.git/target/xtensa/translate.c:1052
#6  0x00005555556448df in gen_intermediate_code (env=0x55555603a890, 
tb=0x7fffccc7de00) at 
/home/alex/lsrc/qemu/qemu.git/target/xtensa/translate.c:3214
#7  0x00005555555e383b in tb_gen_code (address@hidden, address@hidden, 
address@hidden, address@hidden, cflags=<optimised out>, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/translate-all.c:1288
#8  0x00005555555e5e41 in cpu_exec (tb_exit=0, last_tb=<optimised out>, 
cpu=0x38011) at /home/alex/lsrc/qemu/qemu.git/cpu-exec.c:370
#9  0x00005555555e5e41 in cpu_exec (address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/cpu-exec.c:685
#10 0x0000555555611898 in tcg_cpu_exec (cpu=0x555556032600) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1254
#11 0x0000555555611bd4 in qemu_tcg_rr_cpu_thread_fn (arg=<optimised out>) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1350
#12 0x00007fffdf6606ba in start_thread (arg=0x7fffcca12700) at 
pthread_create.c:333
#13 0x00007fffdf39682d in clone () at 
../sysdeps/unix/sysv/linux/x86_64/clone.S:109
Continuing.

Thread 3 "qemu-system-xte" hit Breakpoint 2, cpu_restore_state (address@hidden, 
address@hidden) at /home/alex/lsrc/qemu/qemu.git/translate-all.c:338
338             qemu_log_mask(LOG_UNIMP, "Attempt to resolve CPU state @ 0x%" 
PRIxPTR
#0  0x00005555555e3712 in cpu_restore_state (address@hidden, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/translate-all.c:338
#1  0x000055555564cb38 in tlb_fill (address@hidden, address@hidden, 
address@hidden, address@hidden, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/target/xtensa/op_helper.c:73
#2  0x000055555562d604 in helper_ret_ldb_cmmu (address@hidden, addr=4308992, 
oi=<optimised out>, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/softmmu_template.h:127
#3  0x00005555556448df in gen_intermediate_code (retaddr=0, ptr=<optimised 
out>, env=0x55555603a890) at 
/home/alex/lsrc/qemu/qemu.git/include/exec/cpu_ldst_template.h:102
#4  0x00005555556448df in gen_intermediate_code (ptr=<optimised out>, 
env=0x55555603a890) at 
/home/alex/lsrc/qemu/qemu.git/include/exec/cpu_ldst_template.h:114
#5  0x00005555556448df in gen_intermediate_code (dc=0x7fffcca0f4d0, 
env=0x55555603a890) at 
/home/alex/lsrc/qemu/qemu.git/target/xtensa/translate.c:1052
#6  0x00005555556448df in gen_intermediate_code (env=0x55555603a890, 
tb=0x7fffccd0b1b0) at 
/home/alex/lsrc/qemu/qemu.git/target/xtensa/translate.c:3214
#7  0x00005555555e383b in tb_gen_code (address@hidden, address@hidden, 
address@hidden, address@hidden, cflags=<optimised out>, address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/translate-all.c:1288
#8  0x00005555555e5e41 in cpu_exec (tb_exit=0, last_tb=<optimised out>, 
cpu=0x38011) at /home/alex/lsrc/qemu/qemu.git/cpu-exec.c:370
#9  0x00005555555e5e41 in cpu_exec (address@hidden) at 
/home/alex/lsrc/qemu/qemu.git/cpu-exec.c:685
#10 0x0000555555611898 in tcg_cpu_exec (cpu=0x555556032600) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1254
#11 0x0000555555611bd4 in qemu_tcg_rr_cpu_thread_fn (arg=<optimised out>) at 
/home/alex/lsrc/qemu/qemu.git/cpus.c:1350
#12 0x00007fffdf6606ba in start_thread (arg=0x7fffcca12700) at 
pthread_create.c:333
#13 0x00007fffdf39682d in clone () at 
../sysdeps/unix/sysv/linux/x86_64/clone.S:109


> Is this in fact linux-user, not softmmu, as you imply from tlb_fill?
> Because handle_cpu_signal will in fact pass a genuine non-zero retaddr
> for the SIGSEGV resulting from a cpu_ld*_code from a non-mapped
> address.

I think that is another call chain that might trip us up. Peter
mentioned he'd hit it. This one is definitely softmmu.

> So... for linux-user I think the qemu_log is unnecessary -- that's
> just the way things are.  For softmmu, I don't know what to think
> except that we shouldn't have gotten here.

I kinda agree but all SoftMMU targets have this potential path. Having
saif that I haven't seen it hit ARM, maybe because we take care not to
cross a page boundary?

I agree it would be better to handle fetching code bytes without this
potential for breakage but that would be a much bigger change and need
more testing this close to rc0.

>
>
> r~


--
Alex Bennée