Agreed that this is possibly not optimal, but this patch simply returns the behavior to what it was before "i2c: tegra: Support atomic transfers", fixing the overt issue.
Design fixes can be considered later, in a non-stable patch.
Mikko
On 1/11/21 4:31 PM, David Laight wrote:
From: Mikko Perttunen
Sent: 11 January 2021 13:56
Upon a communication error, the interrupt handler can call tegra_i2c_disable_packet_mode. This causes a sleeping poll to happen unless the current transaction was marked atomic. Since tegra_i2c_disable_packet_mode is only called from the interrupt path, make it use atomic waiting always.
Spin-waiting in an ISR for anything that it makes sense to do a sleep-wait for at other times is badly broken design.
David
Registered Address Lakeside, Bramley Road, Mount Farm, Milton Keynes, MK1 1PT, UK Registration No: 1397386 (Wales)