openfirmware November 2012

openfirmware@openfirmware.info

3 participants
66 discussions

[commit] r3430 - cpu/arm/olpc/4.0
by repository service 15 Nov '12

15 Nov '12

Author: rsmith Date: Thu Nov 15 21:02:44 2012 New Revision: 3430 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3430 Log: OLPC XO-4 - EC Code 0.3.04 Modified: cpu/arm/olpc/4.0/ec-version.fth Modified: cpu/arm/olpc/4.0/ec-version.fth ============================================================================== --- cpu/arm/olpc/4.0/ec-version.fth Thu Nov 15 15:51:28 2012 (r3429) +++ cpu/arm/olpc/4.0/ec-version.fth Thu Nov 15 21:02:44 2012 (r3430) @@ -1,6 +1,6 @@ \ The EC microcode macro: EC_PLATFORM cl4 -macro: EC_VERSION 7_0_3_02 +macro: EC_VERSION 7_0_3_04 \ Alternate command for getting EC microcode, for testing new versions. \ Temporarily uncomment the line and modify the path as necessary

1 0

[commit] r3429 - cpu/arm/mmp2
by repository service 15 Nov '12

15 Nov '12

Author: pgf Date: Thu Nov 15 15:51:28 2012 New Revision: 3429 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3429 Log: fix typo in disabling of static branch prediction Modified: cpu/arm/mmp2/initmmu.fth Modified: cpu/arm/mmp2/initmmu.fth ============================================================================== --- cpu/arm/mmp2/initmmu.fth Wed Nov 14 22:54:24 2012 (r3428) +++ cpu/arm/mmp2/initmmu.fth Thu Nov 15 15:51:28 2012 (r3429) @@ -92,9 +92,9 @@ \ but we found this workaround in the qseven Linux kernel source. \ It fixes a problem we were seeing with Python (trac #12164) so \ I am including it in OFW just in case. - mcr p15,1,r0,cr15,cr1,1 \ Workaround for PJ4B Errata 6409 + mrc p15,1,r0,cr15,cr1,1 \ Workaround for PJ4B Errata 6409 bic r0,r0,#0x04 \ Disable static branch prediction - mrc p15,1,r0,cr15,cr1,1 + mcr p15,1,r0,cr15,cr1,1 [then] mrc p15,0,r0,cr1,cr0,0 \ Read control register

1 0

[commit] r3428 - cpu/arm/olpc/4.0
by repository service 14 Nov '12

14 Nov '12

Author: wmb Date: Wed Nov 14 22:54:24 2012 New Revision: 3428 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3428 Log: Q7B06 Modified: cpu/arm/olpc/4.0/fw-version.fth Modified: cpu/arm/olpc/4.0/fw-version.fth ============================================================================== --- cpu/arm/olpc/4.0/fw-version.fth Wed Nov 14 22:51:30 2012 (r3427) +++ cpu/arm/olpc/4.0/fw-version.fth Wed Nov 14 22:54:24 2012 (r3428) @@ -1,7 +1,7 @@ \ The overall firmware revision macro: FW_PREFIX Q7 macro: FW_MAJOR B -macro: FW_MINOR 05 +macro: FW_MINOR 06 \ Create a 2-character build/fw-suffix file to personalize your test builds " fw-suffix" $file-exists? [if]

1 0

[commit] r3427 - in cpu/arm: mmp2 mmp3
by repository service 14 Nov '12

14 Nov '12

Author: wmb Date: Wed Nov 14 22:51:30 2012 New Revision: 3427 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3427 Log: OLPC ARM MMP3 - Fixed suspend (it didn't sleep fully in Q7B05) by undoing the svn 3410 change in pcr! , and finished moving the wakeup configuration code out of the go-to-sleep code. Modified: cpu/arm/mmp2/rtc.fth cpu/arm/mmp2/wakeups.fth cpu/arm/mmp3/dramrecal.fth Modified: cpu/arm/mmp2/rtc.fth ============================================================================== --- cpu/arm/mmp2/rtc.fth Wed Nov 14 08:47:41 2012 (r3426) +++ cpu/arm/mmp2/rtc.fth Wed Nov 14 22:51:30 2012 (r3427) @@ -20,7 +20,12 @@ : enable-rtc ( -- ) h# 81 0 apbc! ; \ Turn on the clock for the internal RTC : enable-rtc-wakeup ( -- ) - h# 2.0010 h# 104c +mpmu io-set + h# 2.0010 h# 104c +mpmu io-set \ Unmask wakeup 4 (10) and RTC (20000) in PMUM_WUCRM_PJ + + \ Disabling a wakeup decoder in either PCR register disables that wakeup, + \ so we must clear the bit in both registers to allow it. + h# 4.0000 h# 1000 +mpmu io-clr \ Unmask wakeup 4 decoder in PMUM_PCR_PJ + h# 4.0000 h# 0 +mpmu io-clr \ Unmask wakeup 4 decoder in PMUM_PCR_SP ; : soc-rtc@ ( offset -- value ) h# 01.0000 + io@ ; : soc-rtc! ( value offset -- value ) h# 01.0000 + io! ; Modified: cpu/arm/mmp2/wakeups.fth ============================================================================== --- cpu/arm/mmp2/wakeups.fth Wed Nov 14 08:47:41 2012 (r3426) +++ cpu/arm/mmp2/wakeups.fth Wed Nov 14 22:51:30 2012 (r3427) @@ -1,6 +1,29 @@ \ See license at end of file purpose: Setup keyboard wakeups and display the state of the wakeup machinery +0 [if] +: pcr-clr ( mask -- ) dup h# 00 mpmu-clr h# 1000 mpmu-clr ; + +create wakeup-masks \ Bits in MPMU+0 and MPMU+1000, indexed by wakeup port number + h# 0080.0000 , h# 0040.0000 , h# 0020.0000 , h# 0010.0000 , + h# 0004.0000 , h# 0002.0000 , h# 0001.0000 , h# 0000.8000 , + +: port>bit ( port# -- ) wakeup-masks swap na+ @ ; + +: wucrm-set ( mask -- ) dup h# 4c mpmu-set h# 104c mpmu-set ; + +\ We enable the wakeup port in both the SP and PJ registers, thus avoiding +\ any ambiguity about whether the bits are ORed or ANDed +: wakeup-port-on ( port# -- ) + dup port>bit pcr-clr + 1 swap lshift wucrm-set +; +: rtc-wakeup-on ( -- ) + h# 2.0000 wucrm-set + 4 wakeup-port-on +; +[then] + \ XXX we might need to set GPIOs 71 and 160 (ps2 clocks), and perhaps the dat lines too, \ for non-sleep-mode control - or maybe for sleep mode control as inputs. \ We also may need to enable falling edge detects. Modified: cpu/arm/mmp3/dramrecal.fth ============================================================================== --- cpu/arm/mmp3/dramrecal.fth Wed Nov 14 08:47:41 2012 (r3426) +++ cpu/arm/mmp3/dramrecal.fth Wed Nov 14 22:51:30 2012 (r3427) @@ -128,6 +128,12 @@ : cc4-set ( mask -- ) h# 248 pmua-set ; : cc4-clr ( mask -- ) h# 248 pmua-clr ; +\ PXA2128_Registers_Manual_revF.pdf says to always write 0 to bits [14:0] +\ of MPMU+0x1000, but, empirically, that prevents deep sleep. Apparently +\ bits [14:13] are legacy bits like in MPMU+0x0 +: pcr! ( value -- ) dup h# 00 mpmu! h# 1000 mpmu! ; +: pcr-set ( mask -- ) dup h# 00 mpmu-set h# 1000 mpmu-set ; + 0 [if] : pj-c1 ( -- ) h# 62 idle-cfg-clr \ light sleep after WFI @@ -152,26 +158,15 @@ ; [then] -create wakeup-masks \ Bits in MPMU+0 and MPMU+1000, indexed by wakeup port number - h# 0080.0000 , h# 0040.0000 , h# 0020.0000 , h# 0010.0000 , - h# 0004.0000 , h# 0002.0000 , h# 0001.0000 , h# 0000.8000 , - -: port>bit ( port# -- ) wakeup-masks swap na+ @ ; - -: pcr! ( value -- ) dup h# 00 mpmu! h# 7fff invert and h# 1000 mpmu! ; -: pcr-set ( mask -- ) dup h# 00 mpmu-set h# 1000 mpmu-set ; -: pcr-clr ( mask -- ) dup h# 00 mpmu-clr h# 1000 mpmu-clr ; -: wucrm-set ( mask -- ) dup h# 4c mpmu-set h# 104c mpmu-set ; +0 value apcr +: deep-sleep-on ( -- ) + h# 1000 mpmu@ to apcr -\ We enable the wakeup port in both the SP and PJ registers, thus avoiding -\ any ambiguity about whether the bits are ORed or ANDed -: wakeup-port-on ( port# -- ) - dup port>bit pcr-clr - 1 swap lshift wucrm-set + \ Wakeup ports will be handler at a higher level + h# be086000 pcr-set \ AXISD, SPLEN, SPSD, DDRCORSD, APBSD, RSVD, VCXOSD, MSASLPEN, UDR_POWER_OFF_EN \ Step 11 ; -: deep-sleep-on ( -- ) - \ Start with all wakeup ports masked off, enable specific ones later - h# beffe000 pcr! \ AXISD, SPLEN, SPSD, DDRCORSD, APBSD, RSVD, VCXOSD, MSASLPEN, UDR_POWER_OFF_EN \ Step 11 +: deep-sleep-off ( -- ) + apcr h# 1000 mpmu! ; \ D2_L2_PWD 462 @@ -183,10 +178,6 @@ idle-cfg@ h# 0def.fc1d and ( mask kept ) or idle-cfg! ( ) ; -: rtc-wakeup-on ( -- ) - h# 2.0000 wucrm-set - 4 wakeup-port-on -; code outer-flush-all ( -- ) c; code flush-cache-all @@ -251,16 +242,12 @@ 1 h# 2190 icu! \ ICU_GBL_IRQ6_MSK ; -0 value apcr : setup-sleep-state ( -- ) sp-c1-on make-ddr3-recal-tables - h# 1000 mpmu@ to apcr - deep-sleep-on -\ rtc-wakeup-on \ Unnecessary; alarm-in-3 does it h# 462 set-idle \ D2_L2_PWD @@ -270,8 +257,6 @@ \ I don't think we need this because L2 is off \ h# 8000 cc4-set \ workaround: keep SL2 power on -\ wakeup-irqs-on \ Unnecessary; alarm-in-3 does it - global-irqs-off flush-cache-all @@ -281,15 +266,13 @@ \ mp-off ; : restore-run-state ( -- ) - wakeup-irqs-off - \ fw-on \ mp-on \ I don't think we need this because L2 is off \ h# 8000 cc4-clr \ workaround: keep SL2 power on - apcr h# 1000 mpmu! + deep-sleep-off h# 8000.0000 idle-cfg-set \ Workaround: restore AT clock

1 0

[commit] r3426 - cpu/arm/mmp2
by repository service 14 Nov '12

14 Nov '12

Author: wmb Date: Wed Nov 14 08:47:41 2012 New Revision: 3426 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3426 Log: OLPC ARM MMP3 - turn off static branch prediction because it causes erratic results. Modified: cpu/arm/mmp2/initmmu.fth Modified: cpu/arm/mmp2/initmmu.fth ============================================================================== --- cpu/arm/mmp2/initmmu.fth Wed Nov 14 08:41:30 2012 (r3425) +++ cpu/arm/mmp2/initmmu.fth Wed Nov 14 08:47:41 2012 (r3426) @@ -87,6 +87,16 @@ mcr p15, 0, r0, cr7, cr5, 6 \ Flush branch target cache +[ifdef] mmp3 + \ This erratum is not listed in PJ4B-MP.ARMv7-ErrataRev-18.pdf + \ but we found this workaround in the qseven Linux kernel source. + \ It fixes a problem we were seeing with Python (trac #12164) so + \ I am including it in OFW just in case. + mcr p15,1,r0,cr15,cr1,1 \ Workaround for PJ4B Errata 6409 + bic r0,r0,#0x04 \ Disable static branch prediction + mrc p15,1,r0,cr15,cr1,1 +[then] + mrc p15,0,r0,cr1,cr0,0 \ Read control register orr r0,r0,#0x1000 \ ICache on orr r0,r0,#0x0800 \ Branch prediction on

1 0

[commit] r3425 - cpu/arm/mmp3
by repository service 14 Nov '12

14 Nov '12

Author: wmb Date: Wed Nov 14 08:41:30 2012 New Revision: 3425 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3425 Log: OLPC ARM HDMI driver - added hdmi-present? based on GPIO. Modified: cpu/arm/mmp3/hdmi.fth Modified: cpu/arm/mmp3/hdmi.fth ============================================================================== --- cpu/arm/mmp3/hdmi.fth Wed Nov 14 08:31:04 2012 (r3424) +++ cpu/arm/mmp3/hdmi.fth Wed Nov 14 08:41:30 2012 (r3425) @@ -1,6 +1,8 @@ \ See license at end of file purpose: MMP3 HDMI driver +: hdmi-present? ( -- flag ) hdmi-hp-det-gpio# gpio-pin@ 0= ; + : +hdmi h# 20bc00 + ; : hdmi! +hdmi io! ; : hdmi@ +hdmi io@ ;

1 0

[commit] r3424 - cpu/arm/olpc cpu/x86/pc/olpc dev/geode/ac97 dev/hdaudio
by repository service 14 Nov '12

14 Nov '12

Author: wmb Date: Wed Nov 14 08:31:04 2012 New Revision: 3424 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3424 Log: OLPC ARM audio test - workaround channel-swapping problem (trac #12096). Also added some experimental code. Modified: cpu/arm/olpc/sound.fth cpu/x86/pc/olpc/plot.fth dev/geode/ac97/selftest.fth dev/hdaudio/noiseburst.fth dev/hdaudio/test.fth Modified: cpu/arm/olpc/sound.fth ============================================================================== --- cpu/arm/olpc/sound.fth Wed Nov 14 08:11:21 2012 (r3423) +++ cpu/arm/olpc/sound.fth Wed Nov 14 08:31:04 2012 (r3424) @@ -219,6 +219,7 @@ ; : master-rx ( -- ) h# 0c sspa@ h# 8004.0001 or h# 0c sspa! ; \ Master, on : slave-rx ( -- ) h# 0c sspa@ h# 8000.0001 or h# 0c sspa! ; \ Slave, on +: flush-rx ( -- ) h# 0c sspa@ h# 8000.0004 or h# 0c sspa! ; \ Hit the flush bit : disable-sspa-rx ( -- ) h# 0c sspa@ h# 8000.0004 or h# 4.0001 invert and h# 0c sspa! ; : reset-tx ( -- ) h# 8000.0002 h# 8c sspa! ; @@ -260,6 +261,7 @@ ; : master-tx ( -- ) h# 8c sspa@ h# 8004.0001 or h# 8c sspa! ; \ Master, on : slave-tx ( -- ) h# 8c sspa@ h# 8000.0001 or h# 8c sspa! ; \ Slave, on +: flush-tx ( -- ) h# 8c sspa@ h# 8000.0004 or h# 8c sspa! ; \ Hit the flush bit : disable-sspa-tx ( -- ) h# 8c sspa@ h# 8000.0004 or h# 4.0001 invert and h# 8c sspa! ; h# fc0 constant /audio-buf @@ -289,9 +291,14 @@ out-desc h# 30 adma! \ Link to first descriptor out-desc to my-out-desc ; + +0 value use-packmod? + : start-out-ring ( -- ) 1 h# 80 adma! \ Enable DMA completion interrupts - h# 0081.3020 h# 40 adma! \ 16 bits, pack, fetch next, enable, chain, hold dest, inc src + h# 0080.3020 \ 16 bits, fetch next, enable, chain, hold dest, inc src + use-packmod? if h# 1.0000 or then + h# 40 adma! ; : stop-out-ring ( -- ) h# 100000 h# 40 adma! 0 h# 80 adma! ; @@ -304,17 +311,43 @@ : start-in-ring ( -- ) 1 h# 84 adma! \ Enable DMA completion interrupts \ h# 0081.3008 h# 44 adma! \ 16 bits, pack, fetch next, enable, chain, inc dest, hold src - h# 00a1.31c8 h# 44 adma! \ 16 bits, pack, fetch next, enable, chain, burst32, inc dest, hold src + + h# 00a0.31c8 \ 16 bits, fetch next, enable, chain, burst32, inc dest, hold src + use-packmod? if h# 1.0000 or then + h# 44 adma! ; : stop-in-ring ( -- ) h# 100000 h# 44 adma! 0 h# 84 adma! ; +: unpack-move ( src dst packed-len -- ) + 1 rshift 0 ?do ( src dst ) + over i wa+ w@ ( src dst sample ) + over i la+ wa1+ w! ( src dst ) + loop ( src dst ) + 2drop ( ) +; + +: pack-move ( src dst packed-len -- ) + 1 rshift 0 ?do ( src dst ) + over i la+ wa1+ w@ ( src dst sample ) + over i wa+ w! ( src dst ) + loop ( src dst ) + 2drop ( ) +; + : copy-out ( -- ) - my-out-desc >r ( r: desc ) - out-len /audio-buf min ( this-len r: desc ) - dup r@ l! ( this-len r: desc ) - out-adr r@ la1+ l@ third move ( this-len r: desc ) - out-adr over + to out-adr ( this-len r: desc ) - out-len swap - to out-len ( r: desc ) + my-out-desc >r ( r: desc ) + use-packmod? if + out-len /audio-buf min ( this-packed-len r: desc ) + dup r@ l! ( this-packed-len r: desc ) + out-adr r@ la1+ l@ third move ( this-packed-len r: desc ) + else + out-len /audio-buf 2/ min ( this-packed-len r: desc ) + dup 2* r@ l! ( this-packed-len r: desc ) + out-adr r@ la1+ l@ third unpack-move ( this-packed-len r: desc ) + then + + out-adr over + to out-adr ( this-packed-len r: desc ) + out-len swap - to out-len ( r: desc ) out-len if r> 3 la+ l@ to my-out-desc else @@ -323,11 +356,16 @@ ; : copy-in ( -- ) - in-len /audio-buf min ( this-len ) - my-in-desc 2 la+ l@ in-adr third move ( this-len ) - in-adr over + to in-adr ( this-len ) - in-len over - to in-len ( this-len ) - drop ( ) + use-packmod? if + in-len /audio-buf min ( this-packed-len ) + my-in-desc 2 la+ l@ in-adr third move ( this-packed-len ) + else + in-len /audio-buf 2/ min ( this-packed-len ) + my-in-desc 2 la+ l@ in-adr third pack-move ( this-packed-len ) + then + in-adr over + to in-adr ( this-packed-len ) + in-len over - to in-len ( this-packed-len ) + drop ( ) my-in-desc 3 la+ l@ to my-in-desc ; @@ -567,7 +605,17 @@ in-adr0 i la+ w@ in-adr0 i wa+ w! loop ; -: out-in ( out-adr out-len in-adr in-len -- ) +code startit ( sspa-adr -- ) + set r0,#0x80f101f1 + set r1,#0x80f501f1 + str r0,[tos,#0x8c] + str r1,[tos,#0x0c] + pop tos,sp +c; +: startoutin ( -- ) disable-interrupts sspa-base startit enable-interrupts ; +: xstartoutin ( -- ) slave-tx master-rx ; + +: out-in0 ( out-adr out-len in-adr in-len -- ) open-out-in to in-len0 to in-adr0 ( out-adr out-len ) @@ -578,7 +626,7 @@ \ Resetting the clock at this point seems to prevent intermittent channel \ reversal on reception. - audio-clock-on drop ( ) \ This will mess up any frequency settings +\ audio-clock-on drop ( ) \ This will mess up any frequency settings setup-sspa-tx ( ) setup-sspa-rx ( ) @@ -591,7 +639,53 @@ start-in-ring ( ) start-out-ring ( ) + xstartoutin +\ slave-tx ( ) +\ master-rx ( ) \ Now the clock is on + + true to playing? + + begin in-len playing? or while ( ) + in-ready? if copy-in then ( ) + playing? if ?end-playing then ( ) + repeat ( ) + disable-sspa-rx ( ) + disable-sspa-tx ( ) + + stop-in-ring + stop-out-ring + + reset-rx + reset-tx + + close-out-in + mono? if collapse-in then ( ) +; + +: out-in1 ( out-adr out-len in-adr in-len -- ) + open-out-in + + to in-len0 to in-adr0 ( out-adr out-len ) + to out-len to out-adr ( ) + + in-adr0 to in-adr ( ) + in-len0 mono? if 2* then to in-len + + \ Resetting the clock at this point seems to prevent intermittent channel + \ reversal on reception. +\ audio-clock-on drop ( ) \ This will mess up any frequency settings + + make-out-ring ( ) + copy-out ( ) \ Prefill the first Tx buffer + out-len if copy-out then ( ) \ Prefill the second Tx buffer + + setup-sspa-rx ( ) + make-in-ring ( ) + start-in-ring ( ) master-rx ( ) \ Now the clock is on + + start-out-ring ( ) + setup-sspa-tx ( ) slave-tx ( ) true to playing? @@ -613,6 +707,87 @@ mono? if collapse-in then ( ) ; +d# 20 value audio-dly +: out-in2 ( out-adr out-len in-adr in-len -- ) + open-out-in + + to in-len0 to in-adr0 ( out-adr out-len ) + to out-len to out-adr ( ) + + in-adr0 to in-adr ( ) + in-len0 mono? if 2* then to in-len + + \ Resetting the clock at this point seems to prevent intermittent channel + \ reversal on reception. +\ audio-clock-on drop ( ) \ This will mess up any frequency settings + + setup-sspa-tx ( ) + setup-sspa-rx ( ) + + make-in-ring ( ) + make-out-ring ( ) + copy-out ( ) \ Prefill the first Tx buffer + out-len if copy-out then ( ) \ Prefill the second Tx buffer + + xstartoutin + + audio-dly us + flush-rx + flush-tx + start-in-ring ( ) + start-out-ring ( ) + + + true to playing? + + begin in-len playing? or while ( ) + in-ready? if copy-in then ( ) + playing? if ?end-playing then ( ) + repeat ( ) + disable-sspa-rx ( ) + disable-sspa-tx ( ) + + stop-in-ring + stop-out-ring + + reset-rx + reset-tx + + close-out-in + mono? if collapse-in then ( ) +; + +defer out-in ' out-in0 to out-in + +0 [if] +working recipe: only one out-in in test-common +out-in2 +disable-interrupts before audio-clock-on and enable after reset-rx +dly 73 us +read and discard rx fifo depth before flush-rx + +also works with dly 10 us +reliably swaps with dly 1 us +swaps with dly 5 +swap with dly 7 +not swap with dly 8 + +Hmmm, maybe the basic issue is the double out-in in test-common - +Try out-in0 using one out-in +Try out-in2 without read/discard fifo depth using one out-in + +Swap alternates every 10 or so us, flipping at about 8 or 9 mod 10 +An extra 2 samples appear ever 10 or so us, at about 1 or 2 mod 10 + +04 us is stable with swapping +14 us is stable with no swapping + +If you remove flush-rx from fr ( : fr 1c sspa@ foo ! flush-rx ; ) +swapping always happens - 04, 14, 24 us all swap + This is because when you flush-rx with fifo depth 2, 6, a, etc, + that loses an odd number of channel samples +[then] + 0 [if] \ Interactive test words for out-in h# 20000 constant tlen : xb load-base 1meg + ; @@ -730,6 +905,25 @@ fload ${BP}/dev/hdaudio/test.fth warning ! +: wro + wlan-reset-gpio# gpio-dir-in \ So it doesn't fight the cross-wire +; +: input-normal ( -- ) + \ Reconnect pin 36 to WLAN_RESET# + wlan-reset-gpio# af@ 7 invert and wlan-reset-gpio# af! + + \ Reconnect pin 26 back to SSPA1 I2S_DATA_IN + d# 28 af@ 7 invert and 1 or d# 28 af! +; +: input-from-lrclk ( -- ) + \ Move pin 26 from SSPA1 I2S_DATA_IN to SSPA2 I2S_DATA_IN + \ so it doesn't conflict with pin 36. + d# 28 af@ 7 invert and 3 or d# 28 af! + + \ Connect pin 36 to SSPA1 I2S_DATA_IN + wlan-reset-gpio# af@ 7 invert and 2 or wlan-reset-gpio# af! +; + finish-device device-end Modified: cpu/x86/pc/olpc/plot.fth ============================================================================== --- cpu/x86/pc/olpc/plot.fth Wed Nov 14 08:11:21 2012 (r3423) +++ cpu/x86/pc/olpc/plot.fth Wed Nov 14 08:31:04 2012 (r3424) @@ -30,7 +30,10 @@ screen-height wave-height 2* - pitch* frame-buffer-adr + ; -: wave0 ( -- ) screen-height wave-height - ; +0 value wave# +: set-wave# ( n -- ) to wave# ; + +: wave0 ( -- ) screen-height wave# 2* 1+ wave-height * - ; : clear-waveform ( -- ) bg Modified: dev/geode/ac97/selftest.fth ============================================================================== --- dev/geode/ac97/selftest.fth Wed Nov 14 08:11:21 2012 (r3423) +++ dev/geode/ac97/selftest.fth Wed Nov 14 08:31:04 2012 (r3424) @@ -126,6 +126,14 @@ : copy-left-to-right ( adr len -- ) bounds ?do i w@ i wa1+ w! /l +loop ; +: copy-pack ( adr len -- ) + 0 ?do ( adr ) + dup i + w@ ( adr w ) + dup wljoin ( adr l ) + over i 2/ + l! ( adr ) + 8 +loop ( adr ) + drop ( ) +; \ The recording data format is stereo, but usually there is only one mic. \ Depending on the CODEC used, the right channel of the recording is either @@ -140,16 +148,18 @@ d# -3 set-volume play ; +0 value skip-sweep? : selftest ( -- error? ) open 0= if ." Failed to open /audio" cr true exit then wav-test record-len la1+ " dma-alloc" $call-parent to record-base - sweep-test + skip-sweep? 0= if sweep-test then mic-test record-base record-len la1+ " dma-free" $call-parent close false ; +alias st1 selftest \ LICENSE_BEGIN \ Copyright (c) 2007 FirmWorks Modified: dev/hdaudio/noiseburst.fth ============================================================================== --- dev/hdaudio/noiseburst.fth Wed Nov 14 08:11:21 2012 (r3423) +++ dev/hdaudio/noiseburst.fth Wed Nov 14 08:31:04 2012 (r3424) @@ -308,7 +308,7 @@ i <w@ 2 pick - h# 7fff min h# -7fff max i w! i wa1+ <w@ over - h# 7fff min h# -7fff max i wa1+ w! /l +loop ( mean ) - drop ( ) + 2drop ( ) ; : lose-6db ( adr len -- ) bounds ?do ( ) @@ -447,6 +447,9 @@ h# 21000 value /rb \ Mono (stereo for loopback) - 8100 for fixture, 21000 for case, : rb load-base 1meg + ; +: pb+ pb wa1+ ; +: rb+ rb wa1+ ; + : d.. ( -- ) <# # # # # ascii . hold # # # # ascii . hold #s #> type space ; : find-max-mono ( -- ) pb rb /pb 2 / h# 100 - d# 160 d# 120 mono-covar-max .d max-covar d.. @@ -459,10 +462,28 @@ ; : #samples ( -- n ) /pb 4 / h# 100 - ; -: left-range ( -- stereo-adr mono-adr #points ) pb rb #samples ; -: right-range ( -- stereo-adr mono-adr #points ) pb wa1+ rb #samples ; -: left-stereo-range ( -- stereo-adr mono-adr #points ) pb rb #samples ; -: right-stereo-range ( -- stereo-adr mono-adr #points ) pb wa1+ rb wa1+ #samples ; + +defer mono-rb ' rb to mono-rb + +defer left-rb ' rb to left-rb +defer right-rb ' rb+ to right-rb + +defer left-pb ' pb to left-pb +defer right-pb ' pb+ to right-pb + +: swap-lr-pb ( -- ) + ['] pb to right-pb + ['] pb+ to left-pb +; +: unswap-lr-pb ( -- ) + ['] pb to left-pb + ['] pb+ to right-pb +; + +: left-range ( -- stereo-adr mono-adr #points ) left-pb mono-rb #samples ; +: right-range ( -- stereo-adr mono-adr #points ) right-pb mono-rb #samples ; +: left-stereo-range ( -- stereo-adr mono-adr #points ) left-pb left-rb #samples ; +: right-stereo-range ( -- stereo-adr mono-adr #points ) right-pb right-rb #samples ; : fixture-analyze-left ( -- ) left-range d# 146 d# 141 sm-covar-sum dnegate @@ -537,8 +558,8 @@ ; : calc-sm-impulse ( offset -- adr ) \ offset is 0 for left or 2 for right ?alloc-impulse-buf - pb + rb #samples ( adr1 adr2 #samples ) - #impulse-response 0 do + if right-range else left-range then ( adr1 adr2 #samples ) + #impulse-response 0 do ( adr1 adr2 #samples ) 3dup swap i wa+ swap stereo-mono-covar ( adr1 adr2 #samples d.covar ) d# 500,000,000 m/mod nip ( adr1 adr2 #samples n.covar ) impulse-response i wa+ w! ( adr1 adr2 #samples ) @@ -547,9 +568,9 @@ impulse-response ( adr ) ; : calc-stereo-impulse ( offset -- adr ) \ offset is 0 for left or 2 for right - ?alloc-impulse-buf - dup pb + swap rb + #samples ( adr1 adr2 #samples ) - #impulse-response 0 do + ?alloc-impulse-buf ( offset ) + if right-stereo-range else left-stereo-range then ( adr1 adr2 #samples ) + #impulse-response 0 do ( adr1 adr2 #samples ) 3dup swap i la+ swap stereo-covar ( adr1 adr2 #samples d.covar ) d# 50,000,000 m/mod nip ( adr1 adr2 #samples n.covar ) impulse-response i wa+ w! ( adr1 adr2 #samples ) @@ -575,7 +596,12 @@ \ This version puts the tone first into the left channel for \ half the time, then into the right channel for the remainder + +\ This version puts a tone at one frequency in the left channel +\ and a tone at twice that frequency in the right channel + : make-2tones ( adr len freq sample-rate -- ) + 2over erase ( adr len freq sample-rate ) 2dup set-freq ( adr len freq sample-rate ) 3 pick make-cycle drop ( adr len freq sample-rate ) @@ -607,19 +633,17 @@ pb wa1+ /pb -stereo-wmean pb /pb lose-6db pb /pb rb /rb - disable-interrupts +\ disable-interrupts ; : analyze-signal ( -- error? ) - enable-interrupts +\ enable-interrupts rb /rb fix-dc false ( error? ) analyze-left if ( error? ) - ." Left channel failure" cr 1+ then analyze-right if - ." Right channel failure" cr 2+ then ; Modified: dev/hdaudio/test.fth ============================================================================== --- dev/hdaudio/test.fth Wed Nov 14 08:11:21 2012 (r3423) +++ dev/hdaudio/test.fth Wed Nov 14 08:31:04 2012 (r3424) @@ -95,18 +95,25 @@ $call-analyzer ( ) " prepare-signal" $call-analyzer ( pb /pb rb /rb ) \ First shorter run lets the input channel settle - 2over 4 / 2over 4 / out-in ( pb /pb rb /rb ) +\ 2over 4 / 2over 4 / out-in ( pb /pb rb /rb ) out-in ( ) - " analyze-signal" $call-analyzer ( okay? ) + " analyze-signal" $call-analyzer ( error? ) ; +: .test-error ( error? -- error? ) + dup 1 and if ." Left channel failure" cr then + dup 2 and if ." Right channel failure" cr then +; + false value plot? \ Set to true to plot the impulse response, for debugging : plot-impulse ( adr -- ) d# 600 ( adr #samples ) - " 0 set-fg h# ffffffff set-bg single-drawing clear-drawing wave" evaluate - key ascii d = if debug-me then + " 0 set-fg h# ffffffff set-bg single-drawing ( clear-drawing ) wave" evaluate +\ key ascii d = if debug-me then ; +: plot-impulse0 ( adr -- ) 0 " set-wave#" $call-screen plot-impulse ; +: plot-impulse1 ( adr -- ) 1 " set-wave#" $call-screen plot-impulse ; -: test-with-case ( -- ) +: test-with-case ( -- error? ) \ " setup-case" $call-analyzer \ xxx - this needs to use the internal speakers and mic even though the loopback cable is attached true to force-speakers? true to force-internal-mic? @@ -114,11 +121,11 @@ input-test-settings mono output-test-settings case-test-volume set-volume ." Testing internal speakers and microphone" cr - " setup-case" test-common + " setup-case" test-common .test-error false to force-speakers? false to force-internal-mic? plot? if - 0 " calc-sm-impulse" $call-analyzer plot-impulse - 2 " calc-sm-impulse" $call-analyzer plot-impulse + 0 " calc-sm-impulse" $call-analyzer plot-impulse0 + 2 " calc-sm-impulse" $call-analyzer plot-impulse1 then ; : test-with-fixture ( -- error? ) @@ -127,22 +134,41 @@ input-test-settings mono output-test-settings fixture-test-volume set-volume \ -23 prevents obvious visible clipping ." Testing internal speakers and microphone with fixture" cr - " setup-fixture" test-common + " setup-fixture" test-common .test-error false to force-speakers? false to force-internal-mic? plot? if - 0 " calc-sm-impulse" $call-analyzer plot-impulse - 2 " calc-sm-impulse" $call-analyzer plot-impulse + 0 " calc-sm-impulse" $call-analyzer plot-impulse0 + 2 " calc-sm-impulse" $call-analyzer plot-impulse1 then ; +true value allow-swapping? +: ?try-swapped ( error? -- error?' ) + allow-swapping? 0= if exit then ( error? ) + dup if ( error? ) + " swap-lr-pb" $call-analyzer ( error? ) + " analyze-signal" $call-analyzer ( error? swapped-error? ) + " unswap-lr-pb" $call-analyzer ( error? swapped-error? ) + 0= if ( error? ) + \ If swapping left and right "fixes" the problem, we + \ don't report an error. This works around a hard-to-fix + \ random channel-swapping problem with Marvell MMP3. + ." Channel swap!" cr ( error? ) + drop false ( 0 ) + then ( error? ) + then ( error? ) +; : test-with-loopback ( -- error? ) mic-bias-off input-test-settings stereo output-test-settings loopback-test-volume set-volume ." Testing headphone and microphone jacks with loopback cable" cr - " setup-loopback" test-common + " setup-loopback" test-common ( error? ) + ?try-swapped + .test-error + plot? if - 0 " calc-stereo-impulse" $call-analyzer plot-impulse - 2 " calc-stereo-impulse" $call-analyzer plot-impulse + 0 " calc-stereo-impulse" $call-analyzer plot-impulse0 + 2 " calc-stereo-impulse" $call-analyzer plot-impulse1 then ;

1 0

[commit] r3423 - cpu/arm/olpc
by repository service 14 Nov '12

14 Nov '12

Author: wmb Date: Wed Nov 14 08:11:21 2012 New Revision: 3423 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3423 Log: OLPC ARM - accelerometer selftest a) reduced the min-y value to prevent test fallout b) Eliminated the not1g? test because it was failing sometimes. The selftest is now almost worthless because the vendor-recommended acceptance range encompasses pretty much the entire number range. I guess what is being tested is whether the value changes at all (in the right direction) when you invoke the built-in-test feature. Modified: cpu/arm/olpc/accelerometer.fth Modified: cpu/arm/olpc/accelerometer.fth ============================================================================== --- cpu/arm/olpc/accelerometer.fth Wed Nov 14 04:06:40 2012 (r3422) +++ cpu/arm/olpc/accelerometer.fth Wed Nov 14 08:11:21 2012 (r3423) @@ -76,7 +76,7 @@ \ The numbers are in units of "1LSB = 1mg", 1000 unit = 1 gravity. d# 80 value min-x -d# 80 value min-y +d# 20 value min-y d# 80 value min-z d# 1700 value max-x d# 1700 value max-y @@ -132,12 +132,14 @@ delay ( ) average-acceleration@ ( x y z ) +[ifdef] notyet \ Check that the magnitude of the acceleration vector is about 1 G 3dup not1g? if ( x y z ) 3drop ( ) ." Acceleration is not 1 gravity" cr ( ) true exit ( -- error? ) then ( x y z ) +[then] \ Use the device's selftest function to force a change in one direction h# 0a ctl4! ( x y z ) \ High res, Selftest mode 0 @@ -177,7 +179,7 @@ \ The attempt to talk at the old address failed, so we assume the new chip \ Support for new LIS3DHTR chip d# 400,000 to bus-speed - d# 80 to min-x d# 80 to min-y d# 80 to min-z + d# 80 to min-x d# 20 to min-y d# 80 to min-z d# 1700 to max-x d# 1700 to max-y d# 1400 to max-z h# 19 1 reg ['] lis3dhtr-selftest to lis-selftest

1 0

[commit] r3422 - cpu/arm/olpc/1.75
by repository service 14 Nov '12

14 Nov '12

Author: quozl Date: Wed Nov 14 04:06:40 2012 New Revision: 3422 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3422 Log: OLPC XO-1.75 - remove BSL support, was used for prototyping Modified: cpu/arm/olpc/1.75/fw.bth Modified: cpu/arm/olpc/1.75/fw.bth ============================================================================== --- cpu/arm/olpc/1.75/fw.bth Wed Nov 14 00:10:08 2012 (r3421) +++ cpu/arm/olpc/1.75/fw.bth Wed Nov 14 04:06:40 2012 (r3422) @@ -7,9 +7,6 @@ " fw.tag" r/w create-file drop tag-file ! -h# 17000 +io constant bsl-uart-base \ UART2 -h# 30 constant bsl-uart-clock-offset \ APBC_UART2_CLK_RST - fload ${BP}/cpu/arm/olpc/build-fw.fth \ The internal SD card shares the host controller circuitry with

1 0

[commit] r3421 - cpu/arm/olpc
by repository service 14 Nov '12

14 Nov '12

Author: wmb Date: Wed Nov 14 00:10:08 2012 New Revision: 3421 URL: http://tracker.coreboot.org/trac/openfirmware/changeset/3421 Log: OLPC ARM - accelerometer selftest - manufacturing was seeing a small false-negative fallout from the accelerometer test. This change attempts to address it by: a) Changing the scale factor for acceleration components to 1000 units = 1 gravity, per the datasheet definition of "LSB" b) Changing the averaging from 64 samples to 5 samples, per the vendor's recommendation c) Changing the test limits to the vendor's recommended limits (in the new units) d) Fixing a bug in which the first test invocation often failed, due to the sensor initially being in low-resolution mode (so the granularity of the initial average was too coarse). e) Adding a gain test to check that the magnitude of the acceleration vector is about 1 gravity. f) Refactoring the test slightly so that "accelerometer-off" is paired more cleanly with "accelerometer-on" (inside close and open) instead of being all over the place. Modified: cpu/arm/olpc/accelerometer.fth cpu/arm/olpc/roller.fth Modified: cpu/arm/olpc/accelerometer.fth ============================================================================== --- cpu/arm/olpc/accelerometer.fth Thu Nov 8 08:43:41 2012 (r3420) +++ cpu/arm/olpc/accelerometer.fth Wed Nov 14 00:10:08 2012 (r3421) @@ -4,6 +4,8 @@ " accelerometer" name " lis3lv02d" +compatible +true value hi-res? + \ reg is set dynamically by probing to find which chip is present \ This is for the stand-alone accelerometers LIS3DHTR and LIS33DETR @@ -13,17 +15,23 @@ : acc-reg! ( b reg# -- ) " reg-b!" $call-parent ; : ctl1! ( b -- ) h# 20 acc-reg! ; : ctl4! ( b -- ) h# 23 acc-reg! ; -: accelerometer-on ( -- ) h# 47 ctl1! ; +: ctl4@ ( -- b ) h# 23 acc-reg@ ; +: accelerometer-on ( -- ) + h# 47 ctl1! + hi-res? if 8 ctl4! then \ High resolution mode +; : accelerometer-off ( -- ) h# 07 ctl1! ; \ should this be 00? : wext ( b -- n ) dup h# 8000 and if h# ffff0000 or then ; + +\ The scale factor for an acceleration component is 1 gravity = 1000 units. : acceleration@ ( -- x y z ) begin h# 27 acc-reg@ h# 08 and until \ wait for data available h# 0a8 1 6 " bytes-out-in" $call-parent ( xl xh yl yh zl zh ) - bwjoin wext 5 >>a ( xl xh yl yh z ) + bwjoin wext 4 >>a ( xl xh yl yh z ) >r ( xl xh yl yh r: z ) - bwjoin wext 5 >>a ( xl xh y r: z ) + bwjoin wext 4 >>a ( xl xh y r: z ) >r ( xl xh r: z y ) - bwjoin wext 5 >>a ( x r: z y ) + bwjoin wext 4 >>a ( x r: z y ) r> r> ( x y z ) ; @@ -39,15 +47,17 @@ rot r> - ( z1 x3 y3 r: z2 ) rot r> - ( x3 y3 z3 ) ; + \ Averaging a lot of samples reduces the effect of vibration +\ The vendor recommends averaging 5 samples for selftest. : average-acceleration@ ( -- x y z ) acceleration@ ( x y z ) - d# 64 0 do + d# 4 0 do acceleration@ t+ ( x' y' z' ) - loop - rot 6 >>a - rot 6 >>a - rot 6 >>a + loop ( xsum ysum zsum ) + rot 5 / ( ysum zsum x ) + rot 5 / ( zsum x y ) + rot 5 / ( x y z ) ; : delay ( -- ) d# 30 ms ; @@ -61,20 +71,16 @@ accelerometer-off ; -\ The datasheet says the selftest change range is 3 to 32, but we -\ give a little headroom on the high end to allow for external -\ vibration. Typical is supposed to be 19. -\ Empirically, measured maximums were: -\ - Mitch's unit, 32 -\ - James' A3, 41 (on rubber mat on bare ground) -\ - James' A2, 39 - -d# 50 value min-x -d# 50 value min-y -d# 150 value min-z -d# 150 value max-x -d# 150 value max-y -d# 450 value max-z +\ The vendor recommends the following min/max values for LIS3DHTR selftest: +\ X:80..1700 Y:80..1700 Z:80..1400. +\ The numbers are in units of "1LSB = 1mg", 1000 unit = 1 gravity. + +d# 80 value min-x +d# 80 value min-y +d# 80 value min-z +d# 1700 value max-x +d# 1700 value max-y +d# 1400 value max-z : range? ( delta max-delta -- error? ) between 0= ; : error? ( dx dy dz -- error? ) @@ -84,6 +90,20 @@ false ; +\ Lower and upper test limits for the magnitude squared of the acceleration +\ vector when the device is stationary. The scaling is 1000 units = 1 gravity. + +d# 900 dup * constant gsq-min \ 0.9 gravity squared +d# 1100 dup * constant gsq-max \ 1.1 gravity squared + +: xyz>mag-sq ( z y z -- magnitude-squared ) + dup * swap dup * + swap dup * + +; +: not1g? ( x y z -- error? ) + xyz>mag-sq ( magnitude**2 ) + gsq-min gsq-max between 0= +; + : lis33de-selftest ( -- error? ) \ Use the device's selftest function to force a change in one direction delay ( ) @@ -95,7 +115,7 @@ \ because we subtract the new measurement from the old. rot negate -rot ( dx' dy dz ) negate ( dx dy dz' ) - error? if accelerometer-off true exit then + error? if true exit then \ Use the device's selftest function to force a change in the opposite direction accelerometer-on delay ( ) \ Back to normal - STM and STP both off @@ -106,19 +126,26 @@ \ STP applies negative bias to X and Z, but our deltas are inverted \ because we subtract the new measurement from the old. swap negate swap ( dx dy' dz ) - error? if accelerometer-off true exit then - false + error? ; : lis3dhtr-selftest ( -- error? ) - \ Use the device's selftest function to force a change in one direction delay ( ) average-acceleration@ ( x y z ) + + \ Check that the magnitude of the acceleration vector is about 1 G + 3dup not1g? if ( x y z ) + 3drop ( ) + ." Acceleration is not 1 gravity" cr ( ) + true exit ( -- error? ) + then ( x y z ) + + \ Use the device's selftest function to force a change in one direction h# 0a ctl4! ( x y z ) \ High res, Selftest mode 0 delay average-acceleration@ t- ( dx dy dz ) rot negate rot negate rot negate h# 08 ctl4! ( x y z ) \ High res, Normal mode - error? if accelerometer-off true exit then + error? if true exit then \ Use the device's selftest function to force a change in the opposite direction accelerometer-on delay ( ) \ Back to normal - STM and STP both off @@ -127,19 +154,20 @@ delay average-acceleration@ t- ( dx dy dz ) h# 08 ctl4! ( x y z ) \ High res, Normal mode - error? if accelerometer-off true exit then - - accelerometer-off - false + error? ; defer lis-selftest : selftest ( -- error? ) open 0= if true exit then - final-test? if accelerometer-off false exit then + final-test? if + false + else + ." Don't move!" cr + lis-selftest + then - ." Don't move!" cr - lis-selftest + close ; : probe ( -- ) @@ -149,19 +177,21 @@ \ The attempt to talk at the old address failed, so we assume the new chip \ Support for new LIS3DHTR chip d# 400,000 to bus-speed - d# 50 to min-x d# 50 to min-y d# 50 to min-z - d# 150 to max-x d# 150 to max-y d# 450 to max-z + d# 80 to min-x d# 80 to min-y d# 80 to min-z + d# 1700 to max-x d# 1700 to max-y d# 1400 to max-z h# 19 1 reg ['] lis3dhtr-selftest to lis-selftest + true to hi-res? else accelerometer-off \ Something responded to the old address, so we assume it's the old chip \ Support for old LIS33DE chip d# 25,000 to bus-speed - d# 20 to min-x d# 20 to min-y d# 20 to min-z - d# 400 to max-x d# 400 to max-y d# 400 to max-z + d# 40 to min-x d# 40 to min-y d# 40 to min-z + d# 800 to max-x d# 800 to max-y d# 800 to max-z h# 1d 1 reg ['] lis33de-selftest to lis-selftest + false to hi-res? then ; Modified: cpu/arm/olpc/roller.fth ============================================================================== --- cpu/arm/olpc/roller.fth Thu Nov 8 08:43:41 2012 (r3420) +++ cpu/arm/olpc/roller.fth Wed Nov 14 00:10:08 2012 (r3421) @@ -75,7 +75,7 @@ d# 19 d# 20 a/b>fraction value damping -d# 1 d# 80 a/b>fraction value acc-scale +d# 1 d# 160 a/b>fraction value acc-scale d# 20 constant ball-radius d# 40 constant ball-diameter

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

openfirmware November 2012