From: Alan Cox Date: Sun, 25 Nov 2018 15:33:23 +0000 (+0000) Subject: zxdiv: update README X-Git-Url: https://git.ndcode.org/public/gitweb.cgi?a=commitdiff_plain;h=ea6c7412380a3b439204efa03c4dd0e9d31a4bdc;p=FUZIX.git zxdiv: update README --- diff --git a/Kernel/platform-zxdiv/README b/Kernel/platform-zxdiv/README index b9b3930d..43659cac 100644 --- a/Kernel/platform-zxdiv/README +++ b/Kernel/platform-zxdiv/README @@ -63,10 +63,14 @@ around we exchange it with bank 6 as we go. To Do: -- Crash on pre-emption, appears to be a bug in fuse-1.5.4 - Why does ls /bin lose the lowest pixel row of the 'yes' command ? (Seems we lose the bottom line of the left most char - off by one bug in zxvid ?) +- Hand optimized bank exchange and bank to bank copy routines. These are + fairly performance critical due to the limitations of the memory model. + Right now we ldir in one case (could be faster abusing stack), and + the exchange case could be made much much faster using registers + and stack tricks not bounce buffers - Set video space in CODE3 bank to 0 not 0xFF for neatness - Look at what is needed for other suitable interfaces. DIVMMC would in particular be good to support but that means another harder @@ -81,7 +85,8 @@ To Do: RAM (up to 512K). That at the very least would make a kick ass ramdisc/swap device even if the unit lacks the allram switch (May also make sense to hae a build 'the other way up' where we - page apps out of 0000-3FFF copying 4000-7FFF. + page apps out of 0000-3FFF copying 4000-7FFF for allram devices on + a 128K machine - Speccyboot (ENC28J60) - SpectraNet (W5100) @@ -169,3 +174,5 @@ DONE - Check the swap hooks are ok DONE - Turn on swap so we can make progress DONE - Debug crash on swapping DONE - Hangs on pause(). Review p->p_timeout behaviour +DONE - Crash on pre-emption, appears to be a bug in fuse-1.5.4 + (moved code/consts to avoid Fuse bug, filed Fuse bug)