On Thu, Sep 21, 2017 at 09:40:14AM -0500, Larry Finger wrote: > On 09/21/2017 03:07 AM, James Cameron wrote: > >My test kernel "-qb" was write_readback = false in sw.c, with 8-bit > >read of REG_DBI_RDATA, and has been stable for four hours. I'll > >focus on some more testing of this one. It is a surprise. > > > >http://dev.laptop.org/~quozl/z/1dutXk.txt (dmesg) > > > >Observe how REG_DBI_FLAG+0 is briefly seen as 1, which doesn't > >happen with write_readback = true. > > Again, thanks for your efforts. > > At this point, my system has been up over 17 hours without a single > drop. As a result, I will leave the reversion of commit 40b368af4b75 > in place. It seems safer than turning off write_readback. After we > get more testing, that could still be an option. Thanks for the reversion commit, I'll point others to it. My apologies for sloppy work, the test kernel features got swapped! "-qb" above was with write_readback off, and 16-bit read of REG_DBI_RDATA, not 8-bit. Verified with objdump. It has run for 24 hours without a drop. So at conclusion; - the 16-bit read is good with or without write_readback. - the 8-bit read is bad with or without write_readback, and tends to lose connection much quicker without write_readback. Been a pleasure working with you. Back to lurk mode. -- James Cameron http://quozl.netrek.org/
