Hi Martin,
On Mon, Jul 31, 2023 at 11:24 PM Martin K. Petersen martin.petersen@oracle.com wrote:
Hi Ranjan!
Since this patch is a candidate for stable it should be as simple as possible. I don't understand all the complexity introduced to accommodate the new retry_count argument.
static inline u32 -_base_readl_aero(const volatile void __iomem *addr) +_base_readl_aero(const volatile void __iomem *addr, u8 retry_count) { u32 i = 0, ret_val;
do { ret_val = readl(addr); i++;
} while (ret_val == 0 && i < 3);
} while (ret_val == 0 && i < retry_count);_base_readl_aero() is going to return as soon as the register is non-zero. Why is it important that some register reads are only retried 3 times instead of 30? Why can't you just bump the "3" above to "30" and make it a one line change? This HW bug impacts only a few registers. However, for code simplicity, we added retry logic(of 3 retries count) for all registers exposed to the driver.
Now the recommendation is: - Increase retry count from 3 to 30 for only impacted registers. - We also do not want to reduce the retries from 3 to 1 for non-impacted registers to avoid any regression.
Hence, this patch retries upto 30 times for impacted registers and 3 times for non-impacted registers.
Thanks, Ranjan
-- Martin K. Petersen Oracle Linux Engineering