From mboxrd@z Thu Jan 1 00:00:00 1970 From: tony@atomide.com (Tony Lindgren) Date: Fri, 15 Aug 2014 13:28:27 -0700 Subject: [PATCH v7] 8250-core based serial driver for OMAP + DMA In-Reply-To: <53EE5BF8.3010007@linutronix.de> References: <1408124563-31541-1-git-send-email-bigeasy@linutronix.de> <20140815181704.GH17769@csclub.uwaterloo.ca> <53EE5BF8.3010007@linutronix.de> Message-ID: <20140815202826.GC9239@atomide.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org * Sebastian Andrzej Siewior [140815 12:16]: > On 08/15/2014 08:17 PM, Lennart Sorensen wrote: > > > Are you saying that with the new driver you have to respond to the RX > > irq faster than before to avoid overflows? It is not quite clear. > > Yes. The irq fires 46 bytes giving you 16 bytes buffer before overflow > vs 63 bytes buffer the old one had. > > > I do think 40000 interrupts to handle 40000 bytes of date does seem a > > tad inefficient, so dropping to 854 looks a lot nicer. Was the omap > > driver not using the fifo trigger levels at all? > > It configured the trigger levels to 1 for RX and 16 for TX. Hmm that weird RX trigger level is a workaround for lost characters. See commit 0ba5f66836 (tty: serial: OMAP: use a 1-byte RX FIFO threshold in PIO mode :) There's paste test in that commit, I wonder if the 8250 drivers can deal with it any better? Regards, Tony