Re: [PATCH 3/4] linux-user/arm: Handle invalid arm-specific syscalls correctly

[Philippe Mathieu-Daudé]

On 4/21/20 9:44 AM, Edgar E. Iglesias wrote:
> On Mon, Apr 20, 2020 at 10:22:05PM +0100, Peter Maydell wrote:
>> The kernel has different handling for syscalls with invalid
>> numbers that are in the "arm-specific" range 0x9f0000 and up:
>>  * 0x9f0000..0x9f07ff return -ENOSYS if not implemented
>>  * other out of range syscalls cause a SIGILL
>> (see the kernel's arch/arm/kernel/traps.c:arm_syscall())
>>
>> Implement this distinction. (Note that our code doesn't look
>> quite like the kernel's, because we have removed the
>> 0x900000 prefix by this point, whereas the kernel retains
>> it in arm_syscall().)
>>
>> Signed-off-by: Peter Maydell <address@hidden>
>> ---
>>  linux-user/arm/cpu_loop.c | 30 ++++++++++++++++++++++++++----
>>  1 file changed, 26 insertions(+), 4 deletions(-)
>>
>> diff --git a/linux-user/arm/cpu_loop.c b/linux-user/arm/cpu_loop.c
>> index 025887d6b86..f042108b0be 100644
>> --- a/linux-user/arm/cpu_loop.c
>> +++ b/linux-user/arm/cpu_loop.c
>> @@ -332,10 +332,32 @@ void cpu_loop(CPUARMState *env)
>>                              env->regs[0] = cpu_get_tls(env);
>>                              break;
>>                          default:
>> -                            qemu_log_mask(LOG_UNIMP,
>> -                                          "qemu: Unsupported ARM syscall: 
>> 0x%x\n",
>> -                                          n);
>> -                            env->regs[0] = -TARGET_ENOSYS;
>> +                            if (n < 0xf0800) {
>> +                                /*
>> +                                 * Syscalls 0xf0000..0xf07ff (or 0x9f0000..
>> +                                 * 0x9f07ff in OABI numbering) are defined
>> +                                 * to return -ENOSYS rather than raising
>> +                                 * SIGILL. Note that we have already
>> +                                 * removed the 0x900000 prefix.
>> +                                 */
>> +                                qemu_log_mask(LOG_UNIMP,
>> +                                    "qemu: Unsupported ARM syscall: 0x%x\n",
>> +                                              n);
>> +                                env->regs[0] = -TARGET_ENOSYS;
>> +                            } else {
>> +                                /* Otherwise SIGILL */
>> +                                info.si_signo = TARGET_SIGILL;
>> +                                info.si_errno = 0;
>> +                                info.si_code = TARGET_ILL_ILLTRP;
>> +                                info._sifields._sigfault._addr = 
>> env->regs[15];
>> +                                if (env->thumb) {
>> +                                    info._sifields._sigfault._addr -= 2;
>> +                                } else {
>> +                                    info._sifields._sigfault._addr -= 2;
>> +                                }
> 
> 
> Am I missing some detail or are both branches of the if-else doing the
> same thing?

Oops good catch. R-b stands using '-= 4' on 2nd line.

> 
> Cheers,
> Edgar
> 
> 
> 
>> +                                queue_signal(env, info.si_signo,
>> +                                             QEMU_SI_FAULT, &info);
>> +                            }
>>                              break;
>>                          }
>>                      } else {
>> -- 
>> 2.20.1
>>
>>
>