Re: [Qemu-devel] [PATCH/RFC] exec: add cpu_synchronize_state to cpu_memory_rw_debug

[Christian Borntraeger]

On 03/07/2017 04:35 PM, Alex Bennée wrote:
> 
> Christian Borntraeger <address@hidden> writes:
> 
>> I sometimes got "Cannot access memory" when using the x command
>> on the monitor. Turns out that the cpu env did contain stale data
>> (e.g. wrong control register content for page table origin).
>> We must synchronize the state of the CPU before walking the page
>> tables. A similar issues happens for a remote gdb, so lets
>> do the cpu_synchronize_state in cpu_memory_rw_debug.
>>
>> Signed-off-by: Christian Borntraeger <address@hidden>
>> ---
>>  exec.c | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/exec.c b/exec.c
>> index aabb035..e754a03 100644
>> --- a/exec.c
>> +++ b/exec.c
>> @@ -43,6 +43,7 @@
>>  #include "exec/ioport.h"
>>  #include "sysemu/dma.h"
>>  #include "sysemu/numa.h"
>> +#include "sysemu/hw_accel.h"
>>  #include "exec/address-spaces.h"
>>  #include "sysemu/xen-mapcache.h"
>>  #include "trace-root.h"
>> @@ -3309,6 +3310,7 @@ int cpu_memory_rw_debug(CPUState *cpu, target_ulong 
>> addr,
>>      hwaddr phys_addr;
>>      target_ulong page;
>>
>> +    cpu_synchronize_state(cpu);
>>      while (len > 0) {
>>          int asidx;
>>          MemTxAttrs attrs;
> 
> This seems like the wrong place to put it. Would we end up doing a
> potentially expensive sync operations for every byte/word we dump out?

kvm_synchronize_state is clever enough to only sync once. This function mostly
seems to be for debugging stuff (gdb, monitor). Instead of adding the sync
in several places putting it here is a big hammer but it probably fixes all 
places.
I fear that we will forget this in other places. No idea whats right.
 
> Certainly when I was messing around with ARM KVM debug I did the
> synchronise state as we entered the debug handling (e.g.
> gdb_handle_packet/memory_dump)?
> 
> --
> Alex Bennée
>