openfirmware January 2010

openfirmware@openfirmware.info

1 participants
85 discussions

r1666 - ofw/core
by svn＠openfirmware.info 13 Jan '10

13 Jan '10

Author: wmb Date: 2010-01-13 01:32:30 +0100 (Wed, 13 Jan 2010) New Revision: 1666 Modified: ofw/core/osfile.fth Log: /osfile driver - added option for specifying the disk image filename via a defer word instead of in the device specifier. Modified: ofw/core/osfile.fth =================================================================== --- ofw/core/osfile.fth 2010-01-12 23:49:11 UTC (rev 1665) +++ ofw/core/osfile.fth 2010-01-13 00:32:30 UTC (rev 1666) @@ -2,10 +2,28 @@ \ See license at end of file \ Creates a device node named "/osfile", of device-type "block", which -\ accesses an operating system file named by its first argument. +\ accesses an operating system file named by its first argument. That +\ file contains a verbatim disk image. This feature is similar to +\ Linux's "loopback mount" feature. +\ \ Backslash (\) characters in the file name are translated to \ the underlying operating system's directory delimiter. -\ Example: " /osfile:\dev\rfd0" +\ Example: dir /osfile:\dev\rfd0,\boot\test.fth +\ Example: dir /osfile:\home\wmb\fs.img,\boot\test.fth +\ As an alternative to supplying the disk image filename in the device +\ specifier, you can set "osfile$" to return the name. This is +\ helpful for cases where the image filename would interfere with +\ the parsing of later parts of the device specifier. +\ Example: +\ dev /osfile +\ : disk-image$ " /tmp/disk.img" ; +\ ' disk-image$ to osfile$ +\ dend +\ dir /osfile:\boot\test.fth +\ When using the "osfile$" method, you can use either backslash (\) or +\ the system's native delimiter in the disk image filename. When using +\ the device specifier argument method, you must use backslash, because +\ forward slash delimits major device tree components in a device specifier. dev / \ : open true ; : close ; @@ -13,6 +31,7 @@ new-device " osfile" device-name also +defer osfile$ ' null$ to osfile$ 0 instance value file# \ The deblocker converts a block/record-oriented interface to a byte-oriented @@ -56,13 +75,18 @@ ; : $fopen ( adr len -- fd ) $cstr 0 swap 8 syscall 2drop retval ; : open ( -- flag ) - my-args ascii , left-parse-string ( rem-str filename-str ) - 2dup convert/ ( rem-str ) - $fopen to file# ( rem-str ) - file# 0< if 2drop false exit then ( rem-str ) + my-args ( arg$ ) + osfile$ dup 0= if ( arg$ null$ ) + 2drop ( arg$ ) + ascii , left-parse-string ( arg$' img-filename$ ) + then ( arg$ img-filename$ ) - init-deblocker 0= if false exit then ( rem-str ) + 2dup convert/ ( arg$ img-filename$ ) + $fopen to file# ( arg$ ) + file# 0< if 2drop false exit then ( arg$ ) + init-deblocker 0= if false exit then ( arg$ ) + init-label-package dup 0= if ( flag ) deblocker close-package ( false ) file# d# 16 syscall drop ( false )

1 0

r1665 - cpu/x86/pc/olpc/via
by svn＠openfirmware.info 13 Jan '10

13 Jan '10

Author: rsmith Date: 2010-01-13 00:49:11 +0100 (Wed, 13 Jan 2010) New Revision: 1665 Modified: cpu/x86/pc/olpc/via/versions.fth Log: OLPC 1.5: New EC code 1.9.20 Modified: cpu/x86/pc/olpc/via/versions.fth =================================================================== --- cpu/x86/pc/olpc/via/versions.fth 2010-01-12 23:49:08 UTC (rev 1664) +++ cpu/x86/pc/olpc/via/versions.fth 2010-01-12 23:49:11 UTC (rev 1665) @@ -5,7 +5,7 @@ macro: FW_MINOR 26 \ The EC microcode -macro: EC_VERSION 1_9_19 +macro: EC_VERSION 1_9_20 \ Alternate command for getting EC microcode, for testing new versions. \ Temporarily uncomment the line and modify the path as necessary

1 0

r1664 - cpu/x86/pc/olpc/via dev/olpc/kb3700
by svn＠openfirmware.info 13 Jan '10

13 Jan '10

Author: rsmith Date: 2010-01-13 00:49:08 +0100 (Wed, 13 Jan 2010) New Revision: 1664 Modified: cpu/x86/pc/olpc/via/fw.bth dev/olpc/kb3700/ecio.fth Log: OLPC 1.5: Host reboot no longer watchdog reboots the EC. In order to unifiy the reboot code paths 1.5 reboots were changed such that they rebooted the EC as well. This cleared the indexed IO bit so things like flash upgrded didn't need a 2nd reboot. However, We really don't want to be rebooting the EC so much since the battery subsystem does not have any sort of sync to shutdown prior to the reboot. So rather than use cmd66 0xdb cmd there is a new cmd66 0xd7 which will power cycle the host but not reboot the EC. Index IO is now enabled after the EC drops power to the host so it achieves the same behavior as 0xdb. 0xdb is now only for firmware updates. Modified: cpu/x86/pc/olpc/via/fw.bth =================================================================== --- cpu/x86/pc/olpc/via/fw.bth 2010-01-12 23:49:05 UTC (rev 1663) +++ cpu/x86/pc/olpc/via/fw.bth 2010-01-12 23:49:08 UTC (rev 1664) @@ -446,7 +446,7 @@ 2drop 2drop then - ec-reboot \ Tell the EC to reboot us, so the EC can reset too + host-pwr-cycle \ Tell the EC to reboot us \ Prevent "resume from S3" on restart 4 acpi-w@ h# 1c00 invert and 4 acpi-w! Modified: dev/olpc/kb3700/ecio.fth =================================================================== --- dev/olpc/kb3700/ecio.fth 2010-01-12 23:49:05 UTC (rev 1663) +++ dev/olpc/kb3700/ecio.fth 2010-01-12 23:49:08 UTC (rev 1664) @@ -288,6 +288,10 @@ \ EC indexed I/O will come up in enabled state. : ec-reboot ( -- ) h# db ec-cmd66 ; +\ This restarts only the host (no ec reset) but EC index I/O will come up +\ in enable sate +: host-pwr-cycle ( -- ) h# d7 ec-cmd66 ; + : ec-ixio-reboot ( -- ) ['] ec-reboot catch if ." Automatic restart failed. Remove/reinstall the battery and AC." cr

1 0

r1663 - cpu/x86/pc/olpc
by svn＠openfirmware.info 13 Jan '10

13 Jan '10

Author: rsmith Date: 2010-01-13 00:49:05 +0100 (Wed, 13 Jan 2010) New Revision: 1663 Modified: cpu/x86/pc/olpc/rtcwake.fth Log: OLPC: Enhance the autowak functions with extra checks - Often people forget that they need to specify the wakeup time as an argument so try to check for the case when the numbers are obviously wrong. - Run sci-wakeup as well since people forget. Modified: cpu/x86/pc/olpc/rtcwake.fth =================================================================== --- cpu/x86/pc/olpc/rtcwake.fth 2010-01-12 23:49:03 UTC (rev 1662) +++ cpu/x86/pc/olpc/rtcwake.fth 2010-01-12 23:49:05 UTC (rev 1663) @@ -75,7 +75,13 @@ ; : autowack-test ( ms -- ) + dup 0< if ." Delay can't be negative. Did you forget to provide it?" cr + drop exit + then + dup ." Setting autowack delay to " .d ." ms " cr + ." Press x to exit" cr autowack-delay + sci-wakeup autowack-on 0 begin s @@ -88,7 +94,12 @@ \ for testing wakeups from the EC : wackup-test-ec ( ms -- ) + dup 0< if ." Delay can't be negative. Did you forget to provide it?" cr + drop exit + then + dup ." Setting autowack delay to " .d ." ms " cr ." Press x to exit" cr + sci-wakeup d# 3000 autowack-delay \ At small ms delays the host can miss the SCI so this \ is the back up. 0 begin ( ms count )

1 0

r1662 - cpu/x86/pc/olpc/via
by svn＠openfirmware.info 13 Jan '10

13 Jan '10

Author: rsmith Date: 2010-01-13 00:49:03 +0100 (Wed, 13 Jan 2010) New Revision: 1662 Modified: cpu/x86/pc/olpc/via/smt.fth Log: OLPC 1.5: Make smt fw version compare work like in the builds Version compare now checks for the requested version to be > the version on the laptop before tries to do a firmware upgrade Modified: cpu/x86/pc/olpc/via/smt.fth =================================================================== --- cpu/x86/pc/olpc/via/smt.fth 2010-01-12 22:45:35 UTC (rev 1661) +++ cpu/x86/pc/olpc/via/smt.fth 2010-01-12 23:49:03 UTC (rev 1662) @@ -2,6 +2,16 @@ visible +\ Does a base 36 compare on the firmware versions +\ true if $2 <= $1 +: fw-$compare ( $1 $2 -- $2>$1 ) + base @ >r d# 36 base ! ( $1 $2 ) + $number if exit then ( $1 n2 ) + -rot $number if exit then ( n2 n1 ) + pop-base u<= +; + + : wanted-fw$ ( -- $ ) " q3a23" ; @@ -21,7 +31,10 @@ : ?update-firmware ( -- ) \ Exit if the existing firmware and the wanted firmware are the same - fw-version$ wanted-fw$ nocase-$= if exit then + fw-version$ wanted-fw$ fw-$compare if + fw-version$ type ." >= " wanted-fw$ type cr + ." Not updating fimware" cr exit + then ." Updating firmware to version " fw-version$ type cr d# 2000 ms ?enough-power

1 0

r1661 - cpu/x86/pc/olpc/via
by svn＠openfirmware.info 12 Jan '10

12 Jan '10

Author: wmb Date: 2010-01-12 23:45:35 +0100 (Tue, 12 Jan 2010) New Revision: 1661 Modified: cpu/x86/pc/olpc/via/fw.bth Log: OLPC trac 9980 - disable pretty boot during factory test. Modified: cpu/x86/pc/olpc/via/fw.bth =================================================================== --- cpu/x86/pc/olpc/via/fw.bth 2010-01-11 20:43:51 UTC (rev 1660) +++ cpu/x86/pc/olpc/via/fw.bth 2010-01-12 22:45:35 UTC (rev 1661) @@ -582,19 +582,15 @@ \ .built cr ; -: ?factory-boot-sequence ( -- flag ) +: factory-test? ( -- flag ) \ TS is the "test station" tag. Its value is set to "DL" at the \ end of manufacturing test. - " TS" find-tag if ( adr len ) - ?-null " DL" $= 0= ( in-factory? ) - else ( ) - \ Missing TS tag is treated as not factory mode + " TS" find-tag if ( adr len ) + ?-null " SHIP" $= 0= ( in-factory? ) + else ( ) + \ Missing TS tag is treated as not in factory test false - then ( in-factory? ) - if ( ) - " /prober /usb/ethernet disk:\boot\olpc.fth int:\boot\olpc.fth /xp ext:\boot\olpc.fth /wlan" - to boot-device - then ( ) + then ( in-factory? ) ; : debug-interact ( -- ) @@ -650,7 +646,6 @@ no-page ?factory-mode -\ ?factory-boot-sequence disable-user-aborts console-start @@ -663,7 +658,7 @@ read-game-keys - text-off + factory-test? 0= if text-off then " probe-" do-drop-in @@ -688,7 +683,7 @@ ?scan-nand [then] ?fs-update - secure-startup + factory-test? 0= if secure-startup then unblock-exceptions ['] (interrupt-auto-boot?) to interrupt-auto-boot? ?usb-keyboard

1 0

r1660 - cpu/x86/pc/olpc/via
by svn＠openfirmware.info 11 Jan '10

11 Jan '10

Author: pgf Date: 2010-01-11 21:43:51 +0100 (Mon, 11 Jan 2010) New Revision: 1660 Modified: cpu/x86/pc/olpc/via/factory.fth Log: improve messaging surrounding the server protocol. in particular, remove the "pre-check" for the rename target, since it causes extra "SMB Error" messages regarding the (usually) nonexistent file. Modified: cpu/x86/pc/olpc/via/factory.fth =================================================================== --- cpu/x86/pc/olpc/via/factory.fth 2010-01-11 08:06:23 UTC (rev 1659) +++ cpu/x86/pc/olpc/via/factory.fth 2010-01-11 20:43:51 UTC (rev 1660) @@ -69,22 +69,15 @@ : submit-file ( subdir$ -- ) " flush" $call-cifs abort" CIFS flush failed" " do-close" $call-cifs abort" CIFS do-close failed" + ." Creating " 2dup type ." file" cr tempname$ 2swap " %s\\%s" sprintf ( new-name$ ) - - \ Check for preexisting destination file - 2dup 0 " do-open" $call-cifs 0= if ( new-name$ ) - " do-close" $call-cifs drop ( new-name$ ) - ." Error: " type ." already exists on the server" cr ( ) - tempname$ $delete drop ( ) - abort - then ( new-name$ ) - - tempname$ 2swap " $rename" $call-cifs abort" CIFS rename failed" + tempname$ 2swap " $rename" $call-cifs abort" CIFS rename failed. Server program not running?" ; : get-response ( subdir$ -- adr len ) tempname$ 2swap " %s\\%s" sprintf ( response-name$ ) d# 10 0 do ( response-name$ ) d# 1000 ms ( response-name$ ) + ." Looking for response..." cr 2dup 0 " do-open" $call-cifs 0= if ( response-name$ ) 2drop ( ) " size" $call-cifs ( d.size )

1 0

r1659 - in dev/usb2/device: keyboard net serial storage userial wlan
by svn＠openfirmware.info 11 Jan '10

11 Jan '10

Author: wmb Date: 2010-01-11 09:06:23 +0100 (Mon, 11 Jan 2010) New Revision: 1659 Modified: dev/usb2/device/keyboard/kbd.fth dev/usb2/device/net/ethernet.fth dev/usb2/device/serial/uart.fth dev/usb2/device/storage/scsi.fth dev/usb2/device/userial/userial.fth dev/usb2/device/wlan/wlan.fth Log: OLPC trac 9969 - guarded "configuration set-config" with " reset?" $call-parent for all devices. It turns out that if you do set-config unnecessarily for devices where the configuration has already been set, it sometimes fails, and often triggers the problem described in #9423. Modified: dev/usb2/device/keyboard/kbd.fth =================================================================== --- dev/usb2/device/keyboard/kbd.fth 2010-01-10 10:50:45 UTC (rev 1658) +++ dev/usb2/device/keyboard/kbd.fth 2010-01-11 08:06:23 UTC (rev 1659) @@ -326,14 +326,20 @@ ; \ kbd-buf and led-buf must have been allocated -: setup-hardware ( -- ) +: setup-hardware? ( -- error? ) device set-target - configuration set-config if ." Failed to set USB keyboard configuration" cr then + " reset?" $call-parent if + configuration set-config if + ." Failed to set USB keyboard configuration" cr + true exit + then + then set-boot-protocol if ." Failed to set USB keyboard boot protocol" cr then \ Some USB keyboards don't implement set-idle properly, and it's not critical, \ so we suppress the message to avoid confusing the user idle-rate set-idle drop \ if ." Failed to set USB keyboard idle" cr then 0 set-leds + false ; @@ -362,7 +368,10 @@ : open ( -- flag ) kbd-refcount @ if 1 +refcnt true exit then init-kbd-buf - setup-hardware + setup-hardware? if + free-kbd-buf + false exit + then noop \ Add noop so I can patch it before open normal-op? if unlock Modified: dev/usb2/device/net/ethernet.fth =================================================================== --- dev/usb2/device/net/ethernet.fth 2010-01-10 10:50:45 UTC (rev 1658) +++ dev/usb2/device/net/ethernet.fth 2010-01-11 08:06:23 UTC (rev 1659) @@ -211,13 +211,15 @@ : open ( -- ok? ) my-args " debug" $= if debug-on then device set-target - configuration set-config if - ." Failed to set configuration" cr - false exit - then opencount @ 0= if - " reset?" $call-parent if init-nic then + " reset?" $call-parent if + configuration set-config if + ." Failed to set configuration" cr + false exit + then + init-nic + then first-open? if false to first-open? Modified: dev/usb2/device/serial/uart.fth =================================================================== --- dev/usb2/device/serial/uart.fth 2010-01-10 10:50:45 UTC (rev 1658) +++ dev/usb2/device/serial/uart.fth 2010-01-11 08:06:23 UTC (rev 1659) @@ -54,6 +54,14 @@ : open ( -- flag ) device set-target refcount @ 0= if + + " reset?" $call-parent if + configuration set-config if + ." Failed set serial port configuration" cr + false exit + then + then + init-buf inituart rts-dtr-on then Modified: dev/usb2/device/storage/scsi.fth =================================================================== --- dev/usb2/device/storage/scsi.fth 2010-01-10 10:50:45 UTC (rev 1658) +++ dev/usb2/device/storage/scsi.fth 2010-01-11 08:06:23 UTC (rev 1659) @@ -183,8 +183,10 @@ : set-address ( lun -- ) 0 max max-lun min to lun device set-target - configuration set-config if - ." USB storage scsi layer: Failed to set configuration" cr + " reset?" $call-parent if + configuration set-config if + ." USB storage scsi layer: Failed to set configuration" cr + then then ; : set-timeout ( n -- ) bulk-timeout max set-bulk-in-timeout ; Modified: dev/usb2/device/userial/userial.fth =================================================================== --- dev/usb2/device/userial/userial.fth 2010-01-10 10:50:45 UTC (rev 1658) +++ dev/usb2/device/userial/userial.fth 2010-01-11 08:06:23 UTC (rev 1659) @@ -18,9 +18,11 @@ : open ( -- flag ) device set-target - configuration set-config if - ." userial: set-config failed" cr - false exit + " reset?" $call-parent if + configuration set-config if + ." userial: set-config failed" cr + false exit + then then alloc-buffers inbuf h# 10 bulk-in-pipe begin-bulk-in Modified: dev/usb2/device/wlan/wlan.fth =================================================================== --- dev/usb2/device/wlan/wlan.fth 2010-01-10 10:50:45 UTC (rev 1658) +++ dev/usb2/device/wlan/wlan.fth 2010-01-11 08:06:23 UTC (rev 1659) @@ -68,6 +68,12 @@ my-args parse-args set-parent-channel opencount @ 0= if + " reset?" $call-parent if + configuration set-config if + ." Failed to set USB configuration for wireless" cr + false exit + then + then init-buf /inbuf /outbuf setup-bus-io if free-buf false exit then ?load-fw if release-bus-resources free-buf false exit then

1 0

r1658 - ofw/fs/ext2fs
by svn＠openfirmware.info 10 Jan '10

10 Jan '10

Author: wmb Date: 2010-01-10 11:50:45 +0100 (Sun, 10 Jan 2010) New Revision: 1658 Modified: ofw/fs/ext2fs/bitmap.fth Log: Corrected a comment in the last checkin. Modified: ofw/fs/ext2fs/bitmap.fth =================================================================== --- ofw/fs/ext2fs/bitmap.fth 2010-01-10 10:49:32 UTC (rev 1657) +++ ofw/fs/ext2fs/bitmap.fth 2010-01-10 10:50:45 UTC (rev 1658) @@ -164,7 +164,7 @@ : update-gds ( -- ) gd-modified? if \ Copy group descriptors to backup locations - \ XXX if SPARSE_SUPER, then write only to groups 0,1, and powers of 3,5,7 + \ If SPARSE_SUPER, then write only to groups 0,1, and powers of 3,5,7 \ One way to find out, other than checking the numbers, is to inspect \ the block bitmap number in the group descriptor. If it is larger than \ the calculated block number, do the backup.

1 0

r1657 - ofw/fs/ext2fs
by svn＠openfirmware.info 10 Jan '10

10 Jan '10

Author: wmb Date: 2010-01-10 11:49:32 +0100 (Sun, 10 Jan 2010) New Revision: 1657 Modified: ofw/fs/ext2fs/bitmap.fth ofw/fs/ext2fs/dir.fth ofw/fs/ext2fs/sb.fth Log: OLPC trac 9972 - fixed ext2 filesystem corruption that happened in certain cases. The problem was with the sparse superblock feature. You must not write backup copies of the superblock and group descriptors in every group when that feature is enabled. Modified: ofw/fs/ext2fs/bitmap.fth =================================================================== --- ofw/fs/ext2fs/bitmap.fth 2010-01-09 08:12:32 UTC (rev 1656) +++ ofw/fs/ext2fs/bitmap.fth 2010-01-10 10:49:32 UTC (rev 1657) @@ -164,10 +164,19 @@ : update-gds ( -- ) gd-modified? if \ Copy group descriptors to backup locations + \ XXX if SPARSE_SUPER, then write only to groups 0,1, and powers of 3,5,7 + \ One way to find out, other than checking the numbers, is to inspect + \ the block bitmap number in the group descriptor. If it is larger than + \ the calculated block number, do the backup. #groups 1 do - 0 group-desc ( gd0-adr ) - i bpg * 2+ block ( gd0-adr gdn-adr ) - bsize move update ( ) + i bpg * 2+ ( possible-gn ) + dup i block-bitmap < if ( possible-gn ) + block ( gdn-adr ) + 0 group-desc ( gdn-adr gd0-adr ) + swap bsize move update ( ) + else ( possible-gn ) + drop ( ) + then ( ) loop false to gd-modified? update-sb Modified: ofw/fs/ext2fs/dir.fth =================================================================== --- ofw/fs/ext2fs/dir.fth 2010-01-09 08:12:32 UTC (rev 1656) +++ ofw/fs/ext2fs/dir.fth 2010-01-10 10:49:32 UTC (rev 1657) @@ -75,14 +75,13 @@ : +link-count ( increment -- ) link-count + link-count! ; : new-inode ( mode -- inode# ) - alloc-inode set-inode - 0 +i /inode erase ( mode ) + alloc-inode set-inode ( mode ) \ alloc-inode erases the inode file-attr! ( ) time&date >unix-seconds ( time ) dup atime! ( time ) \ set access time dup ctime! ( time ) \ set creation time mtime! ( ) \ set modification time - 1 link-count! ( ) \ set links_count to 1 + 0 link-count! ( ) \ link count will be incremented by new-dirent inode# ( inode# ) ; Modified: ofw/fs/ext2fs/sb.fth =================================================================== --- ofw/fs/ext2fs/sb.fth 2010-01-09 08:12:32 UTC (rev 1656) +++ ofw/fs/ext2fs/sb.fth 2010-01-10 10:49:32 UTC (rev 1657) @@ -28,6 +28,8 @@ : fpg ( -- n ) 9 +sbl ; \ h#2000 frags_per_group : ipg ( -- n ) 10 +sbl ; \ h#790 inodes_per_group : magic ( -- n ) 28 +sbw ; +: state@ ( -- n ) 29 +sbw ; \ for fsck: 0 for dirty, 1 for clean, 2 for known errors +: state! ( n -- ) super-block 29 wa+ short! ; : revlevel ( -- n ) 19 +sbl ; : /inode ( -- n ) revlevel 1 = if 44 +sbw else h# 80 then ; \ : bsize ( -- n )

1 0

← Newer
1
2
3
4
5
6
7
8
9
Older →

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

openfirmware January 2010