Re: [PATCH] tty: mips_ejtag_fdc: Call cpu_relax() in registers polling busy loops

[Zhongqiu Han <quic_zhonhan@quicinc.com>]

On 12/20/2024 3:16 PM, Jiri Slaby wrote:
> On 19. 12. 24, 13:42, Zhongqiu Han wrote:
>> It is considered good practice to call cpu_relax() in busy loops, see
>> Documentation/process/volatile-considered-harmful.rst. This can lower CPU
>> power consumption or yield to a hyperthreaded twin processor, or serve as
>> a compiler barrier. In addition, if something goes wrong in the busy loop
>> at least it can prevent things from getting worse.
>>
>> Signed-off-by: Zhongqiu Han <quic_zhonhan@quicinc.com>
>> ---
>>   drivers/tty/mips_ejtag_fdc.c | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/drivers/tty/mips_ejtag_fdc.c b/drivers/tty/mips_ejtag_fdc.c
>> index afbf7738c7c4..b17ead1e9698 100644
>> --- a/drivers/tty/mips_ejtag_fdc.c
>> +++ b/drivers/tty/mips_ejtag_fdc.c
>> @@ -346,7 +346,7 @@ static void mips_ejtag_fdc_console_write(struct 
>> console *c, const char *s,
>>           /* Busy wait until there's space in fifo */
>>           while (__raw_readl(regs + REG_FDSTAT) & REG_FDSTAT_TXF)
>> -            ;
>> +            cpu_relax();
>>           __raw_writel(word.word, regs + REG_FDTX(c->index));
>>       }
>>   out:
>> @@ -1233,7 +1233,7 @@ static void kgdbfdc_push_one(void)
>>       /* Busy wait until there's space in fifo */
>>       while (__raw_readl(regs + REG_FDSTAT) & REG_FDSTAT_TXF)
>> -        ;
>> +        cpu_relax();
> 
> Can this instead be switched to read_poll_timeout_atomic() or alike? 
> Those already contain cpu_relax(), of course...
> 
> thanks,

Hi Jiri,
Thanks a lot for the review!

yeah, maybe we can consider read_poll_timeout_atomic() or alike.

The implementation of read_poll_timeout_atomic() provides a precise
customization of the address busy read poll behavior by calling udelay()
and cpu_relax(), and using a timeout threshold. However, in this case,
it seems we might not need to customize the poll behavior. Since
udelay() only consumes CPU cycles, perhaps cpu_relax() is sufficient?
And if it times out, we still need to keep retrying until the data is
read. My initial thought was to call cpu_relax() to save power or act as
a memory barrier. As I mentioned before in my email to Greg, certain
MIPS-based architectures, such as Loongson-3, should requirecpu_relax().
Thanks~




-- 
Thx and BRs,
Zhongqiu Han