September 2008 - openfirmware - mail.coreboot.org

r935 - ofw/fs/ext2fs
by svn＠openfirmware.info Sept. 12, 2008

Sept. 12, 2008

Author: wmb Date: 2008-09-13 00:34:01 +0200 (Sat, 13 Sep 2008) New Revision: 935 Modified: ofw/fs/ext2fs/sb.fth Log: Ext2 filesystem - reverted to "read" and "write" method access now that the deblocker has been fixed to eliminate unnecessary reads. Modified: ofw/fs/ext2fs/sb.fth =================================================================== --- ofw/fs/ext2fs/sb.fth 2008-09-12 22:27:47 UTC (rev 934) +++ ofw/fs/ext2fs/sb.fth 2008-09-12 22:34:01 UTC (rev 935) @@ -66,21 +66,16 @@ ; : write-ublocks ( adr len dev-block# -- error? ) -\ ublock um* " seek" $call-parent ?dup if exit then ( adr len ) -\ tuck " write" $call-parent <> - partition-offset + swap ublock 1- + 9 rshift ( adr block# #blocks ) - dup >r " write-blocks" $call-parent r> <> + ublock um* " seek" $call-parent ?dup if exit then ( adr len ) + tuck " write" $call-parent <> ; : put-super-block ( -- error? ) super-block /super-block super-block# write-ublocks ; : read-ublocks ( adr len dev-block# -- error? ) - partition-offset + swap ublock 1- + 9 rshift ( adr block# #blocks ) - dup >r " read-blocks" $call-parent r> <> - -\ ublock um* " seek" $call-parent ?dup if exit then ( adr len ) -\ tuck " read" $call-parent <> + ublock um* " seek" $call-parent ?dup if exit then ( adr len ) + tuck " read" $call-parent <> ; : get-super-block ( -- error? )

1 0

r934 - forth/kernel
by svn＠openfirmware.info Sept. 12, 2008

Sept. 12, 2008

Author: wmb Date: 2008-09-13 00:27:47 +0200 (Sat, 13 Sep 2008) New Revision: 934 Modified: forth/kernel/kernel.fth Log: OLPC trac 8139 - Performance improvement in deblocker - don't read data into buffer if the data is going to be overwritten. Modified: forth/kernel/kernel.fth =================================================================== --- forth/kernel/kernel.fth 2008-09-12 22:23:45 UTC (rev 933) +++ forth/kernel/kernel.fth 2008-09-12 22:27:47 UTC (rev 934) @@ -2061,13 +2061,16 @@ then ; +: align-byte# ( d.byte# -- d.aln-byte# ) fid @ alignop @ execute ; +: byte#-aligned? ( d.byte# -- flag ) 2dup align-byte# d= ; + \ An implementation factor which \ fills the buffer with a block from the current file. The block will \ be chosen so that the file address "d.byte#" is somewhere within that \ block. : fillbuf ( d.byte# -- ) - fid @ alignop @ execute ( d.byte# ) \ Aligns position to a buffer boundary + align-byte# ( d.byte# ) \ Aligns position to a buffer boundary 2dup fstart 2! ( d.byte# ) fid @ seekop @ execute ( ) bfbase @ bflimit @ over - ( adr #bytes-to-read ) @@ -2080,8 +2083,40 @@ \ returns the address within the buffer corresponding to the \ selected position "d.byte#" within the current file. +: bufaddr> ( bufaddr -- d.byte# ) bfbase @ - s>d fstart 2@ d+ ; : >bufaddr ( d.byte# -- bufaddr ) fstart 2@ d- drop bfbase @ + ; +\ This is called from fputs to open up space in the buffer for block-sized +\ chunks, avoiding prefills that would be completely overwritten. +: prefill? ( endaddr curaddr -- endaddr curraddr flag ) + + \ If the current buffer pointer is not block-aligned, must prefill + bfcurrent @ bufaddr> byte#-aligned? 0= if true exit then ( end curr ) + + 2dup - 0 align-byte# drop ( end curr aln-size ) + + \ If the incoming data won't fill a block, must prefill + ?dup 0= if true exit then ( end curr aln-size ) + + \ If there is still space in the buffer, just open it up for copyin + bflimit @ bfend @ - ?dup if ( end curr aln-len buffer-avail ) + min bfend +! false exit + then ( end curr aln-len ) + + \ Save current on stack because ?flushbuf clears it + bfcurrent @ ( end curr aln-len current ) + + \ The buffer is full; clear out its old contents + ?flushbuf ( end curr aln-len ) + + \ Advance the file pointer to the new buffer starting position + bufaddr> fstart 2! ( end curr aln-len ) + + bfbase @ + bflimit @ umin bfend ! ( end curr ) \ Room for new bytes + bfbase @ dup bftop ! bfcurrent ! ( end curr ) \ No valid bytes yet + false +; + \ An implementation factor which \ advances to the next block in the file. This is used when accesses \ to the file are sequential (the most common case). @@ -2352,8 +2387,16 @@ over bfend @ bfbase @ - u>= or if ( d.byte# offset ) \ Not in buffer \ Flush the buffer and get the one containing the desired byte. - drop ?flushbuf 2dup fillbuf ( d.byte# ) - >bufaddr ( bufaddr ) + drop ?flushbuf ( d.byte# ) + 2dup byte#-aligned? if ( d.byte# ) + \ If the new offset is on a block boundary, don't read yet, + \ because the next op could be a large write that fills the buffer. + fstart 2! ( ) + bfbase @ dup bftop ! dup bfend ! ( bufaddr ) + else + 2dup fillbuf ( d.byte# ) + >bufaddr ( bufaddr ) + then ( bufaddr ) else \ The desired byte is already in the buffer. nip nip bfbase @ + ( bufaddr ) @@ -2429,13 +2472,14 @@ \ Writes count bytes from memory starting at "adr" to the current file : fputs ( adr count fd -- ) file @ >r file ! - over + swap ( endaddr startaddr ) - begin copyin 2dup u> - while - \ Here there should be some code to see if there are enough remaining - \ bytes in the request to justify bypassing the file buffer and writing - \ directly from the user's buffer. 'Enough' = more than one file buffer - sync bfcurrent @ shortseek ( endaddr curraddr ) + over + swap ( endaddr startaddr ) + begin copyin 2dup u> while ( endaddr curraddr ) + sync ( endaddr curraddr ) + \ Prefill? tries to avoid unnecessary reads by opening up space + \ in the buffer for chunks that will completely fill a block. + prefill? if ( endaddr curraddr ) + bfcurrent @ shortseek ( endaddr curraddr ) + then repeat 2drop r> file !

1 0

r933 - ofw/inet
by svn＠openfirmware.info Sept. 12, 2008

Sept. 12, 2008

Author: wmb Date: 2008-09-13 00:23:45 +0200 (Sat, 13 Sep 2008) New Revision: 933 Modified: ofw/inet/dhcp.fth Log: OLPC trac 8450 - ask explicitly for several DHCP parameters. Modified: ofw/inet/dhcp.fth =================================================================== --- ofw/inet/dhcp.fth 2008-09-12 20:54:20 UTC (rev 932) +++ ofw/inet/dhcp.fth 2008-09-12 22:23:45 UTC (rev 933) @@ -155,9 +155,22 @@ \ Options common to discover, inform, and request messages : other-parameters ( -- ) set-vendor-class + + + \ Parameter request list + \ h# 01 - 1 Subnet mask + \ h# 03 - 3 Router IP address + \ h# 06 - 6 Name Server IP address + \ h# 0c - 12 Client name + \ h# 0f - 15 Domain name + \ h# 11 - 17 Root path + \ h# 1c - 28 Broadcast IP address + \ h# 2b - 43 Vendor options + \ h# 36 - 54 Server IP address + " "(01 03 06 0c 0f 11 1c 2b 36)" d# 55 +option + \ Later: Add requested IP address if we know it \ Later: Add requested IP lease time if we have a preference - \ Later: Add parameter request list if we care \ Later: Add maximum message size if we should need to ; : use-ip-broadcast ( -- ) broadcast-ip-addr set-dest-ip ; @@ -288,11 +301,37 @@ ; partial-headers -\ For now we'll take the first offer we get. +\ wanted? is the DHCP offer filter. Many environments have multiple DHCP +\ servers, some giving incomplete information. You can set wanted? to +\ require specific characteristics in the offer data. +\ A better approach would be to collect several offers within a time period +\ and choose the best among them, stopping early if a "perfect" offer arrives. + \ The default value of wanted? accepts the first DHCPOFFER that is received defer wanted? ( -- flag ) ' true to wanted? +\ This filter accepts offers that specify a router +: router? ( -- flag ) + 3 find-option dup if nip nip then +; + +\ This filter accepts offers that specify a name server +: ns? ( -- flag ) + 6 find-option dup if nip nip then +; + +: router+ns? ( -- flag ) router? ns? and ; + +: adaptive-wanted? ( -- flag ) + backoff case + d# 8000 of router+ns? exit endof + d# 16000 of router? exit endof + endcase + true +; +' adaptive-wanted? to wanted? \ By default, we accept all DHCP offers + \ This filter rejects offers whose siaddr field is empty, (Microsoft's \ DHCP server doesn't fill in siaddr), since we are hosed if we don't know \ which server to use. @@ -304,9 +343,9 @@ " The DHCP 'siaddr' field is empty" .dhcp-msg then ; -' true to wanted? \ By default, we accept all DHCP offers \ ' (wanted?) to wanted? + \ Another plausible criterion for choosing a particular offer might be: \ If a vendor class identifier is supplied, reject offers that do \ not return that identifier, instead waiting for an offer from a

1 0

r932 - cpu/x86/pc/olpc
by svn＠openfirmware.info Sept. 12, 2008

Sept. 12, 2008

Author: wmb Date: 2008-09-12 22:54:20 +0200 (Fri, 12 Sep 2008) New Revision: 932 Modified: cpu/x86/pc/olpc/lxmsrs.fth Log: OLPC trac 8089 - enable PRNG noise generator in AES unit. Modified: cpu/x86/pc/olpc/lxmsrs.fth =================================================================== --- cpu/x86/pc/olpc/lxmsrs.fth 2008-09-12 00:48:18 UTC (rev 931) +++ cpu/x86/pc/olpc/lxmsrs.fth 2008-09-12 20:54:20 UTC (rev 932) @@ -170,6 +170,7 @@ \ AES \ msr: 5800.2002 00000007.00000007. \ SMI mask - default is all off \ msr: 5800.2004 00000000.00000015. \ Clock gating - default + msr: 5800.2006 00000000.00000503. \ AES control - Random number generator noise, LFSR output \ chipsetinit(nb);

1 0

r931 - dev/olpc/touchpad
by svn＠openfirmware.info Sept. 12, 2008

Sept. 12, 2008

Author: wmb Date: 2008-09-12 02:48:18 +0200 (Fri, 12 Sep 2008) New Revision: 931 Modified: dev/olpc/touchpad/touchpad.fth Log: OLPC touchpad selftest - added support for standard mouse mode for use with Synaptics touchpad. Modified: dev/olpc/touchpad/touchpad.fth =================================================================== --- dev/olpc/touchpad/touchpad.fth 2008-09-11 08:57:08 UTC (rev 930) +++ dev/olpc/touchpad/touchpad.fth 2008-09-12 00:48:18 UTC (rev 931) @@ -19,6 +19,25 @@ ; +h# f800.f800 constant red +h# 07e0.07e0 constant green +h# 001f.001f constant blue +h# ffe0.ffe0 constant yellow +h# f81f.f81f constant magenta +h# 07ff.07ff constant cyan +h# ffff.ffff constant white +h# 0000.0000 constant black + +variable pixcolor + +h# 4 value y-offset +0 value fbadr +0 value maxx +0 value maxy +0 value /line +2 value /pixel + + \ This program depends on the following routines from the \ existing Open Firmware mouse driver: @@ -70,16 +89,16 @@ \ "identify" command; the response (for a mouse-like device) is 0x00 : alps-prefix ( -- ) 3 0 do h# f2 read1 drop loop ; -variable mode +variable mode \ 0 - unknown 1 - GS 2 - PT 3 - mouse \ Ref: 5.2.10 (3) of Hybrid-GP2B-T-1.pdf : gs-only ( -- ) - mode @ 1 = if exit then + mode @ dup 1 = swap 3 = or if exit then alps-prefix mouse1:1 \ f2 f2 f2 e6 1 mode ! ; : pt-only ( -- ) - mode @ 2 = if exit then + mode @ dup 2 = swap 3 = or if exit then alps-prefix mouse2:1 \ f2 f2 f2 e7 2 mode ! ; @@ -88,7 +107,14 @@ \ : simultaneous-mode ( -- ) alps-prefix 2 set-resolution ; \ Put the device into advanced mode and enable it -: start ( -- ) 0 mode ! setup advanced-mode stream-on ; +: start ( -- ) + setup + olpc-touchpad? if + 0 mode ! advanced-mode stream-on + else + remote-mode 3 mode ! + then +; \ I have been unable to get this to work. The response is always \ 64 0 <something>, which doesn't agree with the spec. @@ -237,7 +263,7 @@ h# ff of decode-gs true exit endof \ h# eb of decode-simultaneous exit endof h# aa of - 0 d# 28 at-xy red-screen white-letters + 0 d# 26 at-xy red-screen white-letters ." Unexpected touchpad reset" white-screen black-letters cr @@ -256,10 +282,30 @@ again ; +variable mouse-x +variable mouse-y + +: clipx ( delta -- x ) mouse-x @ + 0 max maxx min dup mouse-x ! ; +: clipy ( delta -- y ) mouse-y @ + 0 max maxy min dup mouse-y ! ; + \ Try to receive a GS-format packet. If one arrives within \ 20 milliseconds, return true and the decoded information. \ Otherwise return false. : pad? ( -- false | x y z tap? true ) + mode @ 3 = if + poll-event if ( dx dy buttons ) + >r ( dx dy ) + swap clipx swap negate clipy 0 ( x y z ) + 0 r@ 1 and or ( x y z tap ) + r> 4 and 0<> 2 and or ( x y z tap' ) + 8 or ( x y z tap' ) + true + else + false + then + exit + then + poll-touchpad 0= if false exit then gx @ gy @ gz @ taps @ 2 lshift switches @ or true @@ -295,31 +341,13 @@ key? until ; -h# f800.f800 constant red -h# 07e0.07e0 constant green -h# 001f.001f constant blue -h# ffe0.ffe0 constant yellow -h# f81f.f81f constant magenta -h# 07ff.07ff constant cyan -h# ffff.ffff constant white -h# 0000.0000 constant black - -variable pixcolor - -h# 4 value y-offset -0 value fbadr -0 value maxx -0 value maxy -0 value /line -2 value /pixel - : button ( color x -- ) maxy d# 50 - d# 200 d# 30 " fill-rectangle" $call-screen ; : background ( -- ) fbadr maxy 2+ /line * erase - 0 d# 30 at-xy ." Touchpad test. Both buttons clears screen. Type a key to exit" cr - 0 d# 20 at-xy ." Pressure: " + 0 d# 27 at-xy ." Touchpad test. Both buttons clears screen. Type a key to exit" cr + mode @ 3 <> if 0 d# 20 at-xy ." Pressure: " then ; : track-init ( -- ) screen-ih package( @@ -330,7 +358,13 @@ : show-up ( x y z -- ) 3drop d# 10 d# 20 at-xy ." UP " ; -: show-pressure ( -- ) push-decimal d# 10 d# 20 at-xy 3 u.r pop-base ; +: show-pressure ( z -- ) + mode @ 3 = if + drop + else + push-decimal d# 10 d# 20 at-xy 3 u.r pop-base + then +; : dot ( x y -- ) y-offset + maxy min /line * ( x line-adr )

1 0

r930 - cpu/x86/pc/olpc
by svn＠openfirmware.info Sept. 11, 2008

Sept. 11, 2008

Author: wmb Date: 2008-09-11 10:57:08 +0200 (Thu, 11 Sep 2008) New Revision: 930 Modified: cpu/x86/pc/olpc/addrs.fth cpu/x86/pc/olpc/gxearly.fth cpu/x86/pc/olpc/gxmsrs.fth cpu/x86/pc/olpc/probemem.fth cpu/x86/pc/olpc/vsapci.fth Log: OLPC trac 8426 - fixed B2 bricking by E-series firmware. The problem was caused by incorrect MSR setting for the memory that's mapped high for OFW's use. The MSRs were set correctly for LX CPUs, but not for GXs. Modified: cpu/x86/pc/olpc/addrs.fth =================================================================== --- cpu/x86/pc/olpc/addrs.fth 2008-09-09 23:12:37 UTC (rev 929) +++ cpu/x86/pc/olpc/addrs.fth 2008-09-11 08:57:08 UTC (rev 930) @@ -17,8 +17,9 @@ dropin-base h# 20 + constant ResetBase \ Location of "reset" dropin in ROM -h# ec0.0000 constant fw-pa +h# ec0.0000 value fw-pa \ Changed in probemem.fth h# 20.0000 constant /fw-ram +h# 40.0000 constant /fw-area [then] [ifdef] linuxbios-loaded Modified: cpu/x86/pc/olpc/gxearly.fth =================================================================== --- cpu/x86/pc/olpc/gxearly.fth 2008-09-09 23:12:37 UTC (rev 929) +++ cpu/x86/pc/olpc/gxearly.fth 2008-09-11 08:57:08 UTC (rev 930) @@ -39,24 +39,27 @@ \ Some of these don't really have to be set here, and could be \ moved to the big table of MSR values, except that the table \ slammer is dumb and can't handle conditionals. - 20000000.000fff00. 10000020 set-msr \ memory - 0..fffff + 20000000.000fff80. 10000020 set-msr \ memory - 0..7ffff + 20000000.080fffe0. 10000026 set-msr \ memory - 80000..9ffff (a0000..bffff is for VGA) + 20000000.0c0fffc0. 10000027 set-msr \ memory - c0000..fffff + ffbff000.ff800100. 1817 set-msr \ Region config - OFW area cacheable 1030 port-rl 4 bitand 0<> if \ 128 MiB - 25fff002.1077e000. 1808 set-msr - 2c7be040.400fffe0. 10000026 set-msr - 20000007.7df00100. 10000028 set-msr \ Top of memory - 20a7e0fd.7fffd000. 10000029 set-msr \ Frame buffer - 10075012.00003400. 20000018 set-msr - 20000007.7df00100. 40000029 set-msr \ top of memory. - 077df000.00100130. 50002019 set-msr + 25fff002.10780000. 1808 set-msr + 20000007.7ff00100. 10000028 set-msr \ Range - Top of memory at 07ff.ffff, fbsize + 20a800fd.7fffd000. 10000029 set-msr \ Range Offset - Frame buffer at PA fd00.0000 maps to RAM at 0780.0000, fbsize + 207c00ff.bffff800. 1000002b set-msr \ Range Offset - OFW area ff80.0000 maps to RAM at 0740.0000, fbsize + 10075012.00003400. 20000018 set-msr \ Refresh/SDRAM - DIMM 0 size 128 MB + 20000007.7ff00100. 40000029 set-msr \ Top of memory for DMA - 77ff.0000, fbsize + 077ff000.00100130. 50002019 set-msr \ Memory Region 1 config - top at 77ff.0000, PF, WC, fbsize else \ 256 MiB - 25fff002.10f7e000. 1808 set-msr - 2cfbe040.400fffe0. 10000026 set-msr - 2000000f.7df00100. 10000028 set-msr \ Top of memory - 2127e0fd.7fffd000. 10000029 set-msr \ Frame buffer - 10076013.00003400. 20000018 set-msr - 2000000f.7df00100. 40000029 set-msr \ top of memory. - 0f7df000.00100130. 50002019 set-msr + 25fff002.10f80000. 1808 set-msr + 2000000f.7ff00100. 10000028 set-msr \ Range - Top of memory at 0fff.ffff, fbsize + 212800fd.7fffd000. 10000029 set-msr \ Range Offset - Frame buffer at PA fd00.0000 maps to RAM at 0f80.0000, fbsize + 20fc00ff.bffff800. 1000002b set-msr \ Range Offset - OFW area ff80.0000 maps to RAM at 0f40.0000, fbsize + 10076013.00003400. 20000018 set-msr \ Refresh/SDRAM - DIMM 0 size 128 MB + 2000000f.7ff00100. 40000029 set-msr \ Top of memory for DMA - f7ff.0000, fbsize + 0f7ff000.00100130. 50002019 set-msr \ Memory Region 1 config - top at f7ff.0000, PF, WC, fbsize then \ 20000019 rmsr \ SDRAM timing and mode program Modified: cpu/x86/pc/olpc/gxmsrs.fth =================================================================== --- cpu/x86/pc/olpc/gxmsrs.fth 2008-09-09 23:12:37 UTC (rev 929) +++ cpu/x86/pc/olpc/gxmsrs.fth 2008-09-11 08:57:08 UTC (rev 930) @@ -52,7 +52,7 @@ \ msr: 4000.0027 000000ff.fff00000. \ Unmapped - default \ msr: 4000.0028 000000ff.fff00000. \ Unmapped - default \ msr: 4000.0029 20000007.7df00100. \ 10.0000 - 0f7d.f000 High RAM - Memsize dependent -msr: 4000.002a 200000fd.7fffd000. \ frame buffer - fd00.0000 .. fd7f.ffff, route to GLIU0 +msr: 4000.002a 200000fd.7fffd000. \ frame buffer - fd00.0000 .. fd7f.ffff, route to GLIU0, fbsize \ msr: 4000.002d 20000000.f0000003. \ 000f.0000 - 000f.ffff expansion ROM; XXX - no expansion ROM msr: 4000.0080 00000000.00000001. \ Route coherency snoops from GLIU1 to GLIU0 msr: 4000.0083 00000000.0000ff00. \ Disable SMIs Modified: cpu/x86/pc/olpc/probemem.fth =================================================================== --- cpu/x86/pc/olpc/probemem.fth 2008-09-09 23:12:37 UTC (rev 929) +++ cpu/x86/pc/olpc/probemem.fth 2008-09-11 08:57:08 UTC (rev 930) @@ -35,6 +35,8 @@ : release-range ( start-adr end-adr -- ) over - release ; : probe ( -- ) + system-ram /fw-area - to fw-pa + 0 total-ram reg \ Report extant memory \ Put h# 10.0000-1f.ffff and 28.0000-memsize in pool, Modified: cpu/x86/pc/olpc/vsapci.fth =================================================================== --- cpu/x86/pc/olpc/vsapci.fth 2008-09-09 23:12:37 UTC (rev 929) +++ cpu/x86/pc/olpc/vsapci.fth 2008-09-11 08:57:08 UTC (rev 930) @@ -154,7 +154,7 @@ create gxfb-hdr \ All R/O except cmd/stat and cache line size ' gxfb-cmd-reg token, ' >hdr-value token, - ff000000 , ffffc000 , ffffc000 , ffffc000 , + ff800000 , ffffc000 , ffffc000 , ffffc000 , 0 , 0 , 0 , 0 , 30100b , 2200002 , 3000000 , 8 ,

1 0

r929 - cpu/x86/pc/olpc
by svn＠openfirmware.info Sept. 9, 2008

Sept. 9, 2008

Author: wmb Date: 2008-09-10 01:12:37 +0200 (Wed, 10 Sep 2008) New Revision: 929 Modified: cpu/x86/pc/olpc/versions.fth Log: Q2E17 (Q2E16 was built from svn 927) Modified: cpu/x86/pc/olpc/versions.fth =================================================================== --- cpu/x86/pc/olpc/versions.fth 2008-09-09 22:34:45 UTC (rev 928) +++ cpu/x86/pc/olpc/versions.fth 2008-09-09 23:12:37 UTC (rev 929) @@ -2,7 +2,7 @@ \ The overall firmware revision macro: FW_MAJOR E -macro: FW_MINOR 15 +macro: FW_MINOR 17 \ The EC microcode macro: EC_VERSION e15

1 0

r928 - cpu/x86/pc/olpc
by svn＠openfirmware.info Sept. 9, 2008

Sept. 9, 2008

Author: wmb Date: 2008-09-10 00:34:45 +0200 (Wed, 10 Sep 2008) New Revision: 928 Modified: cpu/x86/pc/olpc/security.fth Log: OLPC trac 8379 - the screen wasn't freezing (pretty boot) in secure mode. Modified: cpu/x86/pc/olpc/security.fth =================================================================== --- cpu/x86/pc/olpc/security.fth 2008-09-09 02:22:38 UTC (rev 927) +++ cpu/x86/pc/olpc/security.fth 2008-09-09 22:34:45 UTC (rev 928) @@ -56,7 +56,7 @@ : show-going ( -- ) background-rgb rgb>565 progress-xy d# 500 d# 100 " fill-rectangle" $call-screen d# 588 d# 638 to icon-xy " bigdot" show-icon - dcon-unfreeze + " vga?" $call-screen 0= if dcon-unfreeze dcon-freeze then ; : show-x ( -- ) " x" show-icon ; : show-sad ( -- )

1 0

r927 - cpu/x86/pc/olpc
by svn＠openfirmware.info Sept. 9, 2008

Sept. 9, 2008

Author: wmb Date: 2008-09-09 04:22:38 +0200 (Tue, 09 Sep 2008) New Revision: 927 Modified: cpu/x86/pc/olpc/mfgdata.fth Log: OLPC - formatting tweaks to 2-column mfg data display Modified: cpu/x86/pc/olpc/mfgdata.fth =================================================================== --- cpu/x86/pc/olpc/mfgdata.fth 2008-09-09 01:55:47 UTC (rev 926) +++ cpu/x86/pc/olpc/mfgdata.fth 2008-09-09 02:22:38 UTC (rev 927) @@ -77,27 +77,31 @@ : $tag-printable? ( adr len -- flag ) ?-null \ Ignore trailing null - bounds ?do i c@ printable? 0= if false unloop exit then loop + bounds ?do i c@ bl h# 7f within 0= if false unloop exit then loop true ; : wrapped-cdump ( adr len -- ) lmargin @ >r rmargin @ >r tabstops @ >r - 4 lmargin ! d# 78 rmargin ! 3 tabstops ! + 4 lmargin ! d# 71 rmargin ! 3 tabstops ! + dup >r + blue-letters bounds ?do 3 .tab i c@ <# u# u# u#> type loop + black-letters + r> d# 10 > if cr then r> tabstops ! r> rmargin ! r> lmargin ! ; : .mfg-data ( -- ) - 0 lmargin ! d# 78 rmargin ! d# 44 tabstops ! + 0 lmargin ! d# 78 rmargin ! d# 42 tabstops ! ??cr mfg-data-top ( adr ) begin another-tag? while ( adr' data$ tname-adr ) over 4 + .tab 2 type 2 spaces ( adr' data$ ) - 2dup $tag-printable? if ?-null type else wrapped-cdump cr then + 2dup $tag-printable? if ?-null type else wrapped-cdump then exit? if drop exit then repeat drop

1 0

r926 - cpu/x86
by svn＠openfirmware.info Sept. 9, 2008

Sept. 9, 2008

Author: wmb Date: 2008-09-09 03:55:47 +0200 (Tue, 09 Sep 2008) New Revision: 926 Modified: cpu/x86/kerncode.fth Log: Added bskip and lskip primitives to speed up bitmap searching for ext2. Modified: cpu/x86/kerncode.fth =================================================================== --- cpu/x86/kerncode.fth 2008-09-08 23:30:08 UTC (rev 925) +++ cpu/x86/kerncode.fth 2008-09-09 01:55:47 UTC (rev 926) @@ -551,6 +551,31 @@ dx di mov c; +\ Skip initial occurrences of bvalue, returning the residual length +code bskip ( adr len bvalue -- residue ) + di dx mov + ax pop \ BX: compare value + cx pop \ CX: Length + di pop \ SI: address + cld repz byte scas + <> if cx inc then + dx di mov + cx push +c; + +\ Skip initial occurrences of lvalue, returning the residual length +code lskip ( adr len lvalue -- residue ) + di dx mov + ax pop \ BX: compare value + cx pop \ CX: Length + 2 # cx shr \ Convert CX to longword count + di pop \ SI: address + cld repz scas + <> if cx inc then + dx di mov + 2 # cx shl cx push +c; + code noop (s -- ) c; code n->l (s n.unsigned -- l ) c;

1 0