ALSA: emu10k1: fixes related to uploading firmware to the E-MU dock
v2 cover ALSA: emu10k1: fixes related to uploading firmware to the E-MU dock
patches 1, 3 & 5 fix the regression(s), patch 4 fixes a related pre-existing problem, patch 6 is a speculative fix on top of patch 4, and patch 2 is logically part of patch 3, just factored out for legibility.
i don't think it really makes sense to revert fbb64eedf5a3 (ALSA: emu10k1: make E-MU dock monitoring interrupt-driven) and re-do things from scratch, as we'd also need to revert the unrelated c960b012ec47 (ALSA: emu10k1: track loss of external clock on E-MU cards) first.
so i'd just fast-track the series as-is, and cherry-pick it to stable.
---
the series (and a bunch of failed experiments) is fully tested by pietro with both rev1 and rev2 1010 cards, a 0202 cardbus card, and their respective docks. he'll get permanent testing credits in the card capability table later on, so i didn't bother adding a tested-by footer to each patch.
---
v2:
- split off non-bugfix part of the series
- add quotes around summaries in Fixes footers
Oswald Buddenhagen (6):
ALSA: emu10k1: fix E-MU card dock presence monitoring
ALSA: emu10k1: factor out snd_emu1010_load_dock_firmware()
ALSA: emu10k1: move the whole GPIO event handling to the workqueue
ALSA: emu10k1: use mutex for E-MU FPGA access locking
ALSA: emu10k1: fix E-MU dock initialization
ALSA: emu10k1: make E-MU FPGA writes potentially more reliable
include/sound/emu10k1.h | 7 +- sound/pci/emu10k1/emu10k1.c | 3 +- sound/pci/emu10k1/emu10k1_main.c | 158 ++++++++++++++++++------------- sound/pci/emu10k1/emumixer.c | 18 +++- sound/pci/emu10k1/emuproc.c | 9 ++ sound/pci/emu10k1/io.c | 51 +++++----- 6 files changed, 142 insertions(+), 104 deletions(-)
Signature · 1 line
2.44.0.701.g2cf7baacf3.dirty
PATCH v2 1/6 ALSA: emu10k1: fix E-MU card dock presence monitoring
While there are two separate IRQ status bits for dock attach and detach, the hardware appears to mix them up more or less randomly, making them useless for tracking what actually happened. It is much safer to check the dock status separately and proceed based on that, as the old polling code did.
Note that the code assumes that only the dock can be hot-plugged - if other option card bits changed, the logic would break.
--- sound/pci/emu10k1/emu10k1_main.c | 17 ++++++++++------- 1 file changed, 10 insertions(+), 7 deletions(-)
sound/pci/emu10k1/emu10k1_main.c +10 -7
@@ -778,6 +778,11 @@ static void emu1010_firmware_work(struct work_struct *work)msleep (10 );/* Unmute all. Default is muted after a firmware load */ snd_emu1010_fpga_write (emu,EMU_HANA_UNMUTE ,EMU_UNMUTE ); + }else if (!(reg &EMU_HANA_OPTION_DOCK_ONLINE )) { +/* Audio Dock removed */ +dev_info (emu->card->dev,"emu1010: Audio Dock detached\n" ); +/* The hardware auto-mutes all, so we unmute again */ +snd_emu1010_fpga_write (emu,EMU_HANA_UNMUTE ,EMU_UNMUTE ); } }
@@ -810,14 +815,12 @@ static void emu1010_interrupt(struct snd_emu10k1 *emu)u32 sts;snd_emu1010_fpga_read (emu,EMU_HANA_IRQ_STATUS , &sts); -if (sts &EMU_HANA_IRQ_DOCK_LOST ) { -/* Audio Dock removed */ -dev_info (emu->card->dev,"emu1010: Audio Dock detached\n" ); -/* The hardware auto-mutes all, so we unmute again */ -snd_emu1010_fpga_write (emu,EMU_HANA_UNMUTE ,EMU_UNMUTE ); - }else if (sts &EMU_HANA_IRQ_DOCK ) { + +// The distinction of the IRQ status bits is unreliable, +// so we dispatch later based on option card status. +if (sts & (EMU_HANA_IRQ_DOCK |EMU_HANA_IRQ_DOCK_LOST ))schedule_work (&emu->emu1010.firmware_work); - } +if (sts &EMU_HANA_IRQ_WCLK_CHANGED )schedule_work (&emu->emu1010.clock_work); }
Signature · 1 line
2.44.0.701.g2cf7baacf3.dirty
PATCH v2 2/6 ALSA: emu10k1: factor out snd_emu1010_load_dock_firmware()
In reply to Oswald BuddenhagenPulled out of the next patch to improve its legibility.
As the function is now available, call it directly from snd_emu10k1_emu1010_init(), thus making the MicroDock firmware loading synchronous - there isn't really a reason not to. Note that this does not affect the AudioDocks of rev1 cards, as these have no independent power supplies, and thus come up only a while after the main card is initialized.
As a drive-by, adjust the priorities of two messages to better reflect their impact.
--- sound/pci/emu10k1/emu10k1_main.c | 66 +++++++++++++++++--------------- 1 file changed, 36 insertions(+), 30 deletions(-)
sound/pci/emu10k1/emu10k1_main.c +36 -30
@@ -732,11 +732,43 @@ static int snd_emu1010_load_firmware(struct snd_emu10k1 *emu, int dock,return snd_emu1010_load_firmware_entry (emu, *fw); } +static void snd_emu1010_load_dock_firmware (struct snd_emu10k1 *emu) +{ +u32 tmp, tmp2; +int err; + +dev_info (emu->card->dev,"emu1010: Loading Audio Dock Firmware\n" ); +/* Return to Audio Dock programming mode */ +snd_emu1010_fpga_write (emu,EMU_HANA_FPGA_CONFIG , +EMU_HANA_FPGA_CONFIG_AUDIODOCK ); + err =snd_emu1010_load_firmware (emu,1 , &emu->dock_fw); +if (err <0 ) +return ; +snd_emu1010_fpga_write (emu,EMU_HANA_FPGA_CONFIG ,0 ); + +snd_emu1010_fpga_read (emu,EMU_HANA_ID , &tmp); +dev_dbg (emu->card->dev,"emu1010: EMU_HANA+DOCK_ID = 0x%x\n" , tmp); +if ((tmp &0x1f ) !=0x15 ) { +/* FPGA failed to be programmed */ +dev_err (emu->card->dev, +"emu1010: Loading Audio Dock Firmware failed, reg = 0x%x\n" , + tmp); +return ; + } +dev_info (emu->card->dev,"emu1010: Audio Dock Firmware loaded\n" ); + +snd_emu1010_fpga_read (emu,EMU_DOCK_MAJOR_REV , &tmp); +snd_emu1010_fpga_read (emu,EMU_DOCK_MINOR_REV , &tmp2); +dev_info (emu->card->dev,"Audio Dock ver: %u.%u\n" , tmp, tmp2); + +/* Allow DLL to settle, to sync clocking between 1010 and Dock */ +msleep (10 ); +} +static void emu1010_firmware_work (struct work_struct *work) {struct snd_emu10k1 *emu; -u32 tmp, tmp2, reg; -int err; +u32 reg; emu = container_of(work, struct snd_emu10k1, emu1010.firmware_work);
@@ -749,33 +781,7 @@ static void emu1010_firmware_work(struct work_struct *work)snd_emu1010_fpga_read (emu,EMU_HANA_OPTION_CARDS , ®);/* OPTIONS: Which cards are attached to the EMU */ if (reg &EMU_HANA_OPTION_DOCK_OFFLINE ) {/* Audio Dock attached */ -/* Return to Audio Dock programming mode */ -dev_info (emu->card->dev, -"emu1010: Loading Audio Dock Firmware\n" ); -snd_emu1010_fpga_write (emu,EMU_HANA_FPGA_CONFIG , -EMU_HANA_FPGA_CONFIG_AUDIODOCK ); - err =snd_emu1010_load_firmware (emu,1 , &emu->dock_fw); -if (err <0 ) -return ; -snd_emu1010_fpga_write (emu,EMU_HANA_FPGA_CONFIG ,0 ); -snd_emu1010_fpga_read (emu,EMU_HANA_ID , &tmp); -dev_info (emu->card->dev, -"emu1010: EMU_HANA+DOCK_ID = 0x%x\n" , tmp); -if ((tmp &0x1f ) !=0x15 ) { -/* FPGA failed to be programmed */ -dev_info (emu->card->dev, -"emu1010: Loading Audio Dock Firmware file failed, reg = 0x%x\n" , - tmp); -return ; - } -dev_info (emu->card->dev, -"emu1010: Audio Dock Firmware loaded\n" ); -snd_emu1010_fpga_read (emu,EMU_DOCK_MAJOR_REV , &tmp); -snd_emu1010_fpga_read (emu,EMU_DOCK_MINOR_REV , &tmp2); -dev_info (emu->card->dev,"Audio Dock ver: %u.%u\n" , tmp, tmp2); -/* Sync clocking between 1010 and Dock */ -/* Allow DLL to settle */ -msleep (10 ); +snd_emu1010_load_dock_firmware (emu);/* Unmute all. Default is muted after a firmware load */ snd_emu1010_fpga_write (emu,EMU_HANA_UNMUTE ,EMU_UNMUTE ); }else if (!(reg &EMU_HANA_OPTION_DOCK_ONLINE )) {
@@ -892,7 +898,7 @@ static int snd_emu10k1_emu1010_init(struct snd_emu10k1 *emu)snd_emu1010_fpga_read (emu,EMU_HANA_OPTION_CARDS , ®);dev_info (emu->card->dev,"emu1010: Card options = 0x%x\n" , reg);if (reg &EMU_HANA_OPTION_DOCK_OFFLINE ) -schedule_work (&emu->emu1010.firmware_work); +snd_emu1010_load_dock_firmware (emu);if (emu->card_capabilities->no_adat) { emu->emu1010.optical_in =0 ;/* IN_SPDIF */ emu->emu1010.optical_out =0 ;/* OUT_SPDIF */
Signature · 1 line
2.44.0.701.g2cf7baacf3.dirty
PATCH v2 3/6 ALSA: emu10k1: move the whole GPIO event handling to the workqueue
In reply to Oswald BuddenhagenThe actual event processing was already done by workqueue items. We can move the event dispatching there as well, rather than doing it already in the interrupt handler callback.
This change has a rather profound "side effect" on the reliability of the FPGA programming: once we enter programming mode, we must not issue any snd_emu1010_fpga_{read,write}() calls until we're done, as these would badly mess up the programming protocol. But exactly that would happen when trying to program the dock, as that triggers GPIO interrupts as a side effect. This is mitigated by deferring the actual interrupt handling, as workqueue items are not re-entrant.
To avoid scheduling the dispatcher on non-events, we now explicitly ignore GPIO IRQs triggered by "uninteresting" pins, which happens a lot as a side effect of calling snd_emu1010_fpga_{read,write}().
--- include/sound/emu10k1.h | 3 +- sound/pci/emu10k1/emu10k1.c | 3 +- sound/pci/emu10k1/emu10k1_main.c | 56 ++++++++++++++++---------------- 3 files changed, 30 insertions(+), 32 deletions(-)
include/sound/emu10k1.h +1 -2
@@ -1684,8 +1684,7 @@ struct snd_emu1010 {unsigned int clock_fallback;unsigned int optical_in;/* 0:SPDIF, 1:ADAT */ unsigned int optical_out;/* 0:SPDIF, 1:ADAT */ -struct work_struct firmware_work; -struct work_struct clock_work; +struct work_struct work; };struct snd_emu10k1 {
sound/pci/emu10k1/emu10k1.c +1 -2
@@ -189,8 +189,7 @@ static int snd_emu10k1_suspend(struct device *dev) emu->suspend =1 ; -cancel_work_sync (&emu->emu1010.firmware_work); -cancel_work_sync (&emu->emu1010.clock_work); +cancel_work_sync (&emu->emu1010.work);snd_ac97_suspend (emu->ac97);
sound/pci/emu10k1/emu10k1_main.c +28 -28
106 lines not on this page · raw message
Signature · 1 line
2.44.0.701.g2cf7baacf3.dirty
PATCH v2 4/6 ALSA: emu10k1: use mutex for E-MU FPGA access locking
In reply to Oswald BuddenhagenThe FPGA access through the GPIO port does not interfere with other sound processor register access, so there is no need to subject it to emu_lock. And after moving all FPGA access out of the interrupt handler, it does not need to be IRQ-safe, either.
What's more, attaching the dock causes a firmware upload, which takes several seconds. We really don't want to disable IRQs for this long, and even less also have someone else spin with IRQs disabled waiting for us.
Therefore, use a mutex for FPGA access locking.
This makes the code somewhat more noisy, as we need to wrap bigger sections into the mutex, as it needs to enclose the spinlocks.
The latter has the "side effect" of fixing dock FPGA programming in a corner case: a really badly timed mixer access right between entering FPGA programming mode and uploading the netlist would mess up the protocol.
--- include/sound/emu10k1.h | 4 +++ sound/pci/emu10k1/emu10k1_main.c | 19 +++++++++--- sound/pci/emu10k1/emumixer.c | 18 ++++++++---- sound/pci/emu10k1/emuproc.c | 9 ++++++ sound/pci/emu10k1/io.c | 50 ++++++++++++++------------------ 5 files changed, 62 insertions(+), 38 deletions(-)
include/sound/emu10k1.h +4 -0
@@ -1685,6 +1685,7 @@ struct snd_emu1010 {unsigned int optical_in;/* 0:SPDIF, 1:ADAT */ unsigned int optical_out;/* 0:SPDIF, 1:ADAT */ struct work_struct work; +struct mutex lock; };struct snd_emu10k1 {
@@ -1833,6 +1834,9 @@ unsigned int snd_emu10k1_ptr20_read(struct snd_emu10k1 * emu, unsigned int reg,void snd_emu10k1_ptr20_write (struct snd_emu10k1 *emu,unsigned int reg,unsigned int chn,unsigned int data);int snd_emu10k1_spi_write (struct snd_emu10k1 * emu,unsigned int data);int snd_emu10k1_i2c_write (struct snd_emu10k1 *emu,u32 reg,u32 value); +static inline void snd_emu1010_fpga_lock (struct snd_emu10k1 *emu) {mutex_lock (&emu->emu1010.lock); }; +static inline void snd_emu1010_fpga_unlock (struct snd_emu10k1 *emu) {mutex_unlock (&emu->emu1010.lock); }; +void snd_emu1010_fpga_write_lock (struct snd_emu10k1 *emu,u32 reg,u32 value);void snd_emu1010_fpga_write (struct snd_emu10k1 *emu,u32 reg,u32 value);void snd_emu1010_fpga_read (struct snd_emu10k1 *emu,u32 reg,u32 *value);void snd_emu1010_fpga_link_dst_src_write (struct snd_emu10k1 *emu,u32 dst,u32 src);
sound/pci/emu10k1/emu10k1_main.c +15 -4
@@ -810,15 +810,19 @@ static void emu1010_work(struct work_struct *work)return ;#endif +snd_emu1010_fpga_lock (emu); +snd_emu1010_fpga_read (emu,EMU_HANA_IRQ_STATUS , &sts);// The distinction of the IRQ status bits is unreliable, // so we dispatch later based on option card status. if (sts & (EMU_HANA_IRQ_DOCK |EMU_HANA_IRQ_DOCK_LOST ))emu1010_dock_event (emu);if (sts &EMU_HANA_IRQ_WCLK_CHANGED )emu1010_clock_event (emu); + +snd_emu1010_fpga_unlock (emu); }static void emu1010_interrupt (struct snd_emu10k1 *emu)
@@ -852,6 +856,8 @@ static int snd_emu10k1_emu1010_init(struct snd_emu10k1 *emu)* Proper init follows in snd_emu10k1_init(). */ outl (HCFG_LOCKSOUNDCACHE |HCFG_LOCKTANKCACHE_MASK , emu->port +HCFG ); +snd_emu1010_fpga_lock (emu); +/* Disable 48Volt power to Audio Dock */ snd_emu1010_fpga_write (emu,EMU_HANA_DOCK_PWR ,0 );
@@ -877,17 +883,18 @@ static int snd_emu10k1_emu1010_init(struct snd_emu10k1 *emu) err =snd_emu1010_load_firmware (emu,0 , &emu->firmware);if (err <0 ) {dev_info (emu->card->dev,"emu1010: Loading Firmware failed\n" ); -return err; + goto fail; }/* ID, should read & 0x7f = 0x55 when FPGA programmed. */ snd_emu1010_fpga_read (emu,EMU_HANA_ID , ®);if ((reg &0x3f ) !=0x15 ) {/* FPGA failed to be programmed */ dev_info (emu->card->dev,"emu1010: Loading Hana Firmware file failed, reg = 0x%x\n" , reg); -return -ENODEV ; + err = -ENODEV ; + goto fail; }dev_info (emu->card->dev,"emu1010: Hana Firmware loaded\n" );
@@ -947,7 +954,9 @@ static int snd_emu10k1_emu1010_init(struct snd_emu10k1 *emu)// so it is safe to simply enable the outputs. snd_emu1010_fpga_write (emu,EMU_HANA_UNMUTE ,EMU_UNMUTE ); -return 0 ; +fail: +snd_emu1010_fpga_unlock (emu); +return err; }/* * Create the EMU10K1 instance
@@ -969,9 +978,10 @@ static void snd_emu10k1_free(struct snd_card *card) }if (emu->card_capabilities->emu_model ==EMU_MODEL_EMU1010 ) {/* Disable 48Volt power to Audio Dock */ -snd_emu1010_fpga_write (emu,EMU_HANA_DOCK_PWR ,0 ); +snd_emu1010_fpga_write_lock (emu,EMU_HANA_DOCK_PWR ,0 ); }cancel_work_sync (&emu->emu1010.work); +mutex_destroy (&emu->emu1010.lock);release_firmware (emu->firmware);release_firmware (emu->dock_fw);snd_util_memhdr_free (emu->memhdr);
@@ -1551,6 +1561,7 @@ int snd_emu10k1_create(struct snd_card *card, emu->synth =NULL ; emu->get_synth_voice =NULL ;INIT_WORK (&emu->emu1010.work, emu1010_work); +mutex_init (&emu->emu1010.lock);/* read revision & serial */ emu->revision = pci->revision;pci_read_config_dword (pci,PCI_SUBSYSTEM_VENDOR_ID , &emu->serial);
sound/pci/emu10k1/emumixer.c +13 -5
@@ -661,7 +661,9 @@ static int snd_emu1010_output_source_put(struct snd_kcontrol *kcontrol, change = (emu->emu1010.output_source[channel] != val);if (change) { emu->emu1010.output_source[channel] = val; +snd_emu1010_fpga_lock (emu);snd_emu1010_output_source_apply (emu, channel, val); +snd_emu1010_fpga_unlock (emu); }return change; }
@@ -705,7 +707,9 @@ static int snd_emu1010_input_source_put(struct snd_kcontrol *kcontrol, change = (emu->emu1010.input_source[channel] != val);if (change) { emu->emu1010.input_source[channel] = val; +snd_emu1010_fpga_lock (emu);snd_emu1010_input_source_apply (emu, channel, val); +snd_emu1010_fpga_unlock (emu); }return change; }
@@ -774,7 +778,7 @@ static int snd_emu1010_adc_pads_put(struct snd_kcontrol *kcontrol, struct snd_ct cache = cache & ~mask; change = (cache != emu->emu1010.adc_pads);if (change) { -snd_emu1010_fpga_write (emu,EMU_HANA_ADC_PADS , cache ); +snd_emu1010_fpga_write_lock (emu,EMU_HANA_ADC_PADS , cache ); emu->emu1010.adc_pads = cache; }
@@ -832,7 +836,7 @@ static int snd_emu1010_dac_pads_put(struct snd_kcontrol *kcontrol, struct snd_ct cache = cache & ~mask; change = (cache != emu->emu1010.dac_pads);if (change) { -snd_emu1010_fpga_write (emu,EMU_HANA_DAC_PADS , cache ); +snd_emu1010_fpga_write_lock (emu,EMU_HANA_DAC_PADS , cache ); emu->emu1010.dac_pads = cache; }
@@ -980,6 +984,7 @@ static int snd_emu1010_clock_source_put(struct snd_kcontrol *kcontrol, val = ucontrol->value.enumerated.item[0 ] ;if (val >= emu_ci->num)return -EINVAL ; +snd_emu1010_fpga_lock (emu);spin_lock_irq (&emu->reg_lock); change = (emu->emu1010.clock_source != val);if (change) {
@@ -996,6 +1001,7 @@ static int snd_emu1010_clock_source_put(struct snd_kcontrol *kcontrol, }else {spin_unlock_irq (&emu->reg_lock); } +snd_emu1010_fpga_unlock (emu);return change; }
@@ -1041,7 +1047,7 @@ static int snd_emu1010_clock_fallback_put(struct snd_kcontrol *kcontrol, change = (emu->emu1010.clock_fallback != val);if (change) { emu->emu1010.clock_fallback = val; -snd_emu1010_fpga_write (emu,EMU_HANA_DEFCLOCK ,1 - val); +snd_emu1010_fpga_write_lock (emu,EMU_HANA_DEFCLOCK ,1 - val); }return change; }
@@ -1093,7 +1099,7 @@ static int snd_emu1010_optical_out_put(struct snd_kcontrol *kcontrol, emu->emu1010.optical_out = val; tmp = (emu->emu1010.optical_in ?EMU_HANA_OPTICAL_IN_ADAT :EMU_HANA_OPTICAL_IN_SPDIF ) | (emu->emu1010.optical_out ?EMU_HANA_OPTICAL_OUT_ADAT :EMU_HANA_OPTICAL_OUT_SPDIF ); -snd_emu1010_fpga_write (emu,EMU_HANA_OPTICAL_TYPE , tmp); +snd_emu1010_fpga_write_lock (emu,EMU_HANA_OPTICAL_TYPE , tmp); }return change; }
@@ -1144,7 +1150,7 @@ static int snd_emu1010_optical_in_put(struct snd_kcontrol *kcontrol, emu->emu1010.optical_in = val; tmp = (emu->emu1010.optical_in ?EMU_HANA_OPTICAL_IN_ADAT :EMU_HANA_OPTICAL_IN_SPDIF ) | (emu->emu1010.optical_out ?EMU_HANA_OPTICAL_OUT_ADAT :EMU_HANA_OPTICAL_OUT_SPDIF ); -snd_emu1010_fpga_write (emu,EMU_HANA_OPTICAL_TYPE , tmp); +snd_emu1010_fpga_write_lock (emu,EMU_HANA_OPTICAL_TYPE , tmp); }return change; }
@@ -2323,7 +2329,9 @@ int snd_emu10k1_mixer(struct snd_emu10k1 *emu,for (i =0 ; i < emu_ri->n_outs; i++) emu->emu1010.output_source[i] =emu1010_map_source (emu_ri, emu_ri->out_dflts[i]); +snd_emu1010_fpga_lock (emu);snd_emu1010_apply_sources (emu); +snd_emu1010_fpga_unlock (emu); kctl = emu->ctl_clock_source =snd_ctl_new1 (&snd_emu1010_clock_source, emu); err =snd_ctl_add (card, kctl);
sound/pci/emu10k1/emuproc.c +9 -0
@@ -165,6 +165,8 @@ static void snd_emu10k1_proc_spdif_read(struct snd_info_entry *entry,u32 value2;if (emu->card_capabilities->emu_model) { +snd_emu1010_fpga_lock (emu); +// This represents the S/PDIF lock status on 0404b, which is // kinda weird and unhelpful, because monitoring it via IRQ is // impractical (one gets an IRQ flood as long as it is desynced).
@@ -197,6 +199,8 @@ static void snd_emu10k1_proc_spdif_read(struct snd_info_entry *entry,snd_iprintf (buffer,"\nS/PDIF mode: %s%s\n" , value &EMU_HANA_SPDIF_MODE_RX_PRO ?"professional" :"consumer" , value &EMU_HANA_SPDIF_MODE_RX_NOCOPY ?", no copy" :"" ); + +snd_emu1010_fpga_unlock (emu); }else {snd_emu10k1_proc_spdif_status (emu, buffer,"CD-ROM S/PDIF In" ,CDCS ,CDSRCS );snd_emu10k1_proc_spdif_status (emu, buffer,"Optical or Coax S/PDIF In" ,GPSCS ,GPSRCS );
@@ -458,6 +462,9 @@ static void snd_emu_proc_emu1010_reg_read(struct snd_info_entry *entry,struct snd_emu10k1 *emu = entry->private_data;u32 value;int i; + +snd_emu1010_fpga_lock (emu); +snd_iprintf (buffer,"EMU1010 Registers:\n\n" );for (i =0 ; i <0x40 ; i+=1 ) {
@@ -496,6 +503,8 @@ static void snd_emu_proc_emu1010_reg_read(struct snd_info_entry *entry,snd_emu_proc_emu1010_link_read (emu, buffer,0x701 ); } } + +snd_emu1010_fpga_unlock (emu); } static void snd_emu_proc_io_reg_read(struct snd_info_entry *entry,
sound/pci/emu10k1/io.c +21 -29
92 lines not on this page · raw message
Signature · 1 line
2.44.0.701.g2cf7baacf3.dirty
PATCH v2 5/6 ALSA: emu10k1: fix E-MU dock initialization
In reply to Oswald BuddenhagenA side effect of making the dock monitoring interrupt-driven was that we'd be very quick to program a freshly connected dock. However, for unclear reasons, the dock does not work when we do that - despite the FPGA netlist upload going just fine. We work around this by adding a delay before programming the dock; for safety, the value is several times as much as was determined empirically.
Note that a badly timed dock hot-plug would have triggered the problem even before the referenced commit - but now it would happen 100% instead of about 3% of the time, thus making it impossible to work around by re-plugging.
--- sound/pci/emu10k1/emu10k1_main.c | 6 ++++++ 1 file changed, 6 insertions(+)
sound/pci/emu10k1/emu10k1_main.c +6 -0
@@ -737,6 +737,12 @@ static void snd_emu1010_load_dock_firmware(struct snd_emu10k1 *emu)u32 tmp, tmp2;int err; +// The docking events clearly arrive prematurely - while the +// Dock's FPGA seems to be successfully programmed, the Dock +// fails to initialize subsequently if we don't give it some +// time to "warm up" here. +msleep (200 ); +dev_info (emu->card->dev,"emu1010: Loading Audio Dock Firmware\n" );/* Return to Audio Dock programming mode */ snd_emu1010_fpga_write(emu, EMU_HANA_FPGA_CONFIG,
Signature · 1 line
2.44.0.701.g2cf7baacf3.dirty
PATCH v2 6/6 ALSA: emu10k1: make E-MU FPGA writes potentially more reliable
In reply to Oswald BuddenhagenWe did not delay after the second strobe signal, so another immediately following access could potentially corrupt the written value.
This is a purely speculative fix with no supporting evidence, but after taking out the spinlocks around the writes, it seems plausible that a modern processor could be actually too fast. Also, it's just cleaner to be consistent.
--- sound/pci/emu10k1/io.c | 1 + 1 file changed, 1 insertion(+)
sound/pci/emu10k1/io.c +1 -0
@@ -285,6 +285,7 @@ static void snd_emu1010_fpga_write_locked(struct snd_emu10k1 *emu, u32 reg, u32outw (value, emu->port +A_GPIO );udelay (10 );outw (value |0x80 , emu->port +A_GPIO );/* High bit clocks the value into the fpga. */ +udelay (10 ); }void snd_emu1010_fpga_write (struct snd_emu10k1 *emu,u32 reg,u32 value)
Signature · 1 line
2.44.0.701.g2cf7baacf3.dirty
Takashi Iwai
AppliedQuoted 19 lines · Oswald Buddenhagen
patches 1, 3 & 5 fix the regression(s), patch 4 fixes a related pre-existing problem, patch 6 is a speculative fix on top of patch 4, and patch 2 is logically part of patch 3, just factored out for legibility.
i don't think it really makes sense to revert
fbb64eedf5a3(ALSA: emu10k1: make E-MU dock monitoring interrupt-driven) and re-do things from scratch, as we'd also need to revert the unrelatedc960b012ec47(ALSA: emu10k1: track loss of external clock on E-MU cards) first.so i'd just fast-track the series as-is, and cherry-pick it to stable.
---
the series (and a bunch of failed experiments) is fully tested by pietro with both rev1 and rev2 1010 cards, a 0202 cardbus card, and their respective docks. he'll get permanent testing credits in the card capability table later on, so i didn't bother adding a tested-by footer to each patch.
Quoted 4 lines
---
v2:
- split off non-bugfix part of the series
- add quotes around summaries in Fixes footers
Applied to for-linus branch. Thanks.
Takashi