qemu-riscv
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [RFC PATCH v2 6/8] cpu: Declare cpu_has_work() in 'sysemu/tcg.h'


From: Philippe Mathieu-Daudé
Subject: Re: [RFC PATCH v2 6/8] cpu: Declare cpu_has_work() in 'sysemu/tcg.h'
Date: Mon, 8 Mar 2021 14:37:19 +0100
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.7.0

On 3/8/21 1:17 PM, Claudio Fontana wrote:
> On 3/4/21 11:23 PM, Philippe Mathieu-Daudé wrote:
>> We can only check if a vCPU has work with TCG.
>> Move the cpu_has_work() prototype to "sysemu/tcg.h".
>>
>> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
>> ---
>> RFC: could another accelerator do that?
>>      can we rename this tcg_vcpu_has_work()?
>> ---
>>  include/hw/core/cpu.h | 16 ----------------
>>  include/sysemu/tcg.h  | 11 +++++++++++
>>  accel/tcg/cpu-exec.c  |  7 +++++++
>>  softmmu/cpus.c        |  1 +
>>  4 files changed, 19 insertions(+), 16 deletions(-)
>>
>> diff --git a/include/hw/core/cpu.h b/include/hw/core/cpu.h
>> index 1376e496a3f..66109bcca35 100644
>> --- a/include/hw/core/cpu.h
>> +++ b/include/hw/core/cpu.h
>> @@ -670,22 +670,6 @@ CPUState *cpu_create(const char *typename);
>>   */
>>  const char *parse_cpu_option(const char *cpu_option);
>>  
>> -/**
>> - * cpu_has_work:
>> - * @cpu: The vCPU to check.
>> - *
>> - * Checks whether the CPU has work to do.
>> - *
>> - * Returns: %true if the CPU has work, %false otherwise.
>> - */
>> -static inline bool cpu_has_work(CPUState *cpu)
>> -{
>> -    CPUClass *cc = CPU_GET_CLASS(cpu);
>> -
>> -    g_assert(cc->has_work);
>> -    return cc->has_work(cpu);
>> -}
>> -
>>  /**
>>   * qemu_cpu_is_self:
>>   * @cpu: The vCPU to check against.
>> diff --git a/include/sysemu/tcg.h b/include/sysemu/tcg.h
>> index c16c13c3c69..3d46b0a7a93 100644
>> --- a/include/sysemu/tcg.h
>> +++ b/include/sysemu/tcg.h
>> @@ -10,6 +10,7 @@
>>  
>>  #ifndef CONFIG_TCG
>>  #define tcg_enabled() 0
>> +#define cpu_has_work(cpu) false
> 
> email errors, so not sure if my previous message went off.

Yes, got it.

> 
> I am not convinced by this approach of _hiding_ the fact that cpu_has_work is 
> TCG-only from the context where it is used.
> 
> When reading the actual code using this, how to get an immediate clue that 
> this is TCG-only?
> 
> I think cpu_has_work should either be renamed to tcg_cpu_has_work(), thus 
> providing the context in the name,
> or it should not be visible/called for non-TCG at all, its call protected in 
> some way (if (tcg_enabled() or ifdef or such..)

If no other accelerator can do that, then I agree. I don't know enough
all accelerators to be sure.

> 
> Ciao,
> 
> Claudio
> 
> 
>>  #else
>>  
>>  void tcg_exec_init(unsigned long tb_size, int splitwx);
>> @@ -26,6 +27,16 @@ extern bool tcg_allowed;
>>  extern bool mttcg_enabled;
>>  #define qemu_tcg_mttcg_enabled() (mttcg_enabled)
>>  
>> +/**
>> + * cpu_has_work:
>> + * @cpu: The vCPU to check.
>> + *
>> + * Checks whether the CPU has work to do.
>> + *
>> + * Returns: %true if the CPU has work, %false otherwise.
>> + */
>> +bool cpu_has_work(CPUState *cpu);
>> +
>>  #endif /* CONFIG_TCG */
>>  
>>  #endif
>> diff --git a/accel/tcg/cpu-exec.c b/accel/tcg/cpu-exec.c
>> index 7e67ade35b9..b9ce36e59e2 100644
>> --- a/accel/tcg/cpu-exec.c
>> +++ b/accel/tcg/cpu-exec.c
>> @@ -447,6 +447,13 @@ static inline TranslationBlock *tb_find(CPUState *cpu,
>>      return tb;
>>  }
>>  
>> +bool cpu_has_work(CPUState *cpu)
>> +{
>> +    CPUClass *cc = CPU_GET_CLASS(cpu);
>> +
>> +    return cc->has_work(cpu);
>> +}
>> +
>>  static inline bool cpu_handle_halt(CPUState *cpu)
>>  {
>>      if (cpu->halted) {
>> diff --git a/softmmu/cpus.c b/softmmu/cpus.c
>> index a7ee431187a..548ab9236f1 100644
>> --- a/softmmu/cpus.c
>> +++ b/softmmu/cpus.c
>> @@ -42,6 +42,7 @@
>>  #include "sysemu/runstate.h"
>>  #include "sysemu/cpu-timers.h"
>>  #include "sysemu/whpx.h"
>> +#include "sysemu/tcg.h"
>>  #include "hw/boards.h"
>>  #include "hw/hw.h"
>>  
>>
> 
> 



reply via email to

[Prev in Thread] Current Thread [Next in Thread]