Re: [PATCH v2 10/17] drm/msm/a6xx: Update HFI definitions

[Akhil P Oommen <akhilpo@oss.qualcomm.com>]

On 3/30/2026 4:45 PM, Konrad Dybcio wrote:
> On 3/27/26 1:13 AM, Akhil P Oommen wrote:
>> Update the HFI definitions to support additional GMU based power
>> features.
>>
>> Signed-off-by: Akhil P Oommen <akhilpo@oss.qualcomm.com>
>> ---
> 
> Whether you want to proceed with bitfields or not:
> 
> Reviewed-by: Konrad Dybcio <konrad.dybcio@oss.qualcomm.com>

Thanks. I still feel it is bitfield layout is a 'bit' complicated.

I did an experiment:

#include <stdio.h>
#include <stdint.h>
#include <stddef.h>

struct packed_u8 {
        uint16_t mitigation_time;
        uint8_t current_rating    : 6;
        uint8_t num_phases        : 6;
        uint8_t path_notification : 1;
        uint8_t extra_features    : 3;
} __packed;


void main() {

        struct packed_u8 data = {
                 .mitigation_time  = 0xffff,
                 .current_rating   = 0x3f,  /* all 6 bits set */
                 .num_phases       = 0x3f,
                 .path_notification = 1,
                 .extra_features   = 0x7,
         };

        printf("Akhil 0x%x\n", *((uint32_t *) &data));
}

The output I got in Kaanapali is: Akhil 0x7f3fffff

This means that the compiler inserted a padding between current_rating
and num_phases.

-Akhil.

> 
> Konrad