| Message ID | 20231211102217.2436294-1-quan@os.amperecomputing.com |
|---|---|
| Headers | show |
| Series | i2c: aspeed: Late ack Tx done irqs and handle coalesced start with stop conditions | expand |
On Mon, 11 Dec 2023 at 20:52, Quan Nguyen <quan@os.amperecomputing.com> wrote: > > This series consists of two patches to handle the below issues observed > when testing with slave mode: > + The coalesced stop condition with the start conditions > + Early ack'ed of Tx done (ACK and NAK) causing "Unexpected Ack on > read request". Looks good. I've reached out to a few people who use slave mode to ask for review and testing on hardware. As long as they don't come back with issues, we should get this merged and backported to stable. Reviewed-by: Joel Stanley <joel@jms.id.au> Cheers, Joel
On Tue, 2023-12-12 at 13:09 +1030, Joel Stanley wrote: > On Mon, 11 Dec 2023 at 20:52, Quan Nguyen <quan@os.amperecomputing.com> wrote: > > > > This series consists of two patches to handle the below issues observed > > when testing with slave mode: > > + The coalesced stop condition with the start conditions > > + Early ack'ed of Tx done (ACK and NAK) causing "Unexpected Ack on > > read request". > > Looks good. I've reached out to a few people who use slave mode to ask > for review and testing on hardware. As long as they don't come back > with issues, we should get this merged and backported to stable. > > Reviewed-by: Joel Stanley <joel@jms.id.au> I've dropped a build with the patches onto an AST2600 EVB and lightly exercised the affected paths using NVMe-MI over MCTP to a Micron 7450. Tested-by: Andrew Jeffery <andrew@codeconstruct.com.au> Andrew
This series consists of two patches to handle the below issues observed when testing with slave mode: + The coalesced stop condition with the start conditions + Early ack'ed of Tx done (ACK and NAK) causing "Unexpected Ack on read request". This series was verified with ast2500 and ast2600. The prior discussion could be found at: https://lore.kernel.org/all/20231208033142.1673232-1-quan@os.amperecomputing.com/ v4: + Switch to use define macro instead of variable [Andrew] + Make the early ack conditionally to avoid unnecessary writel()/readl() [Quan] + Add an extra empty line before the comment in patch 1 [Andrew] v3: + Fix the unconditional write when ack the irqs [Andrew] + Handle the coalesced stop condition with the start conditions [Andrew] + Refactor the code to enhance code readability [Quan] + Revised commit message [Quan] v2: + Split these patches to separate series [Joel] + Added the Fixes lines [Joel] + Fixed multiline comment [Joel] + Refactor irq clearing code [Joel, Guenter] + Revised commit message [Joel, Quan] v1: + These patches are first introduced from this disscusstion https://lore.kernel.org/all/20210519074934.20712-1-quan@os.amperecomputing.com/ Quan Nguyen (2): i2c: aspeed: Handle the coalesced stop conditions with the start conditions. i2c: aspeed: Acknowledge Tx done with and without ACK irq late drivers/i2c/busses/i2c-aspeed.c | 75 +++++++++++++++++++++++---------- 1 file changed, 52 insertions(+), 23 deletions(-)