* [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation @ 2018-05-12 0:05 Calvin Lee 2018-05-12 0:05 ` [Qemu-devel] [PATCH RFC v2 1/2] PC Chipset: Improve serial divisor calculation Calvin Lee ` (2 more replies) 0 siblings, 3 replies; 6+ messages in thread From: Calvin Lee @ 2018-05-12 0:05 UTC (permalink / raw) To: qemu-devel; +Cc: mst, pbonzini, dgilbert, Calvin Lee Hello, While developing a serial implementation for my OS, I found several bugs in QEMU's serial device. I confirmed (by testing on my x64 laptop) that there are several inconsistancies between QEMU and hardware in this regard. For both patches, I used "http://www.sci.muni.cz/docs/pc/serport.txt" for reference. First, QEMU has several errors for setting the UART divisor that are fixed in my first patch. Second, (and more importantly) QEMU does not transmit serial bytes at the correct rate, and this is fixed in my second patch. I have neither contributed to QEMU nor sent patches to a mailing list before, so I marked this patch-series as RFC. I would appreciate any comments you have. Also, I am not very confident in VM migration between versions, so I would appreciate if someone could help me make sure this is correct in my patches. v2: fix build Calvin Lee (2): PC Chipset: Improve serial divisor calculation PC Chipset: Send serial bytes at correct rate hw/char/serial.c | 74 +++++++++++++++++++++++++++++++--------- include/hw/char/serial.h | 2 ++ 2 files changed, 60 insertions(+), 16 deletions(-) -- 2.17.0 ^ permalink raw reply [flat|nested] 6+ messages in thread
* [Qemu-devel] [PATCH RFC v2 1/2] PC Chipset: Improve serial divisor calculation 2018-05-12 0:05 [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation Calvin Lee @ 2018-05-12 0:05 ` Calvin Lee 2018-05-12 0:05 ` [Qemu-devel] [PATCH RFC v2 2/2] PC Chipset: Send serial bytes at correct rate Calvin Lee 2018-07-15 15:57 ` [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation Paolo Bonzini 2 siblings, 0 replies; 6+ messages in thread From: Calvin Lee @ 2018-05-12 0:05 UTC (permalink / raw) To: qemu-devel; +Cc: mst, pbonzini, dgilbert, Calvin Lee This fixes several problems I found in the UART serial implementation. Now all divisor values are allowed, while before divisor values of zero and below the base baud rate were rejected. All changes are in reference to http://www.sci.muni.cz/docs/pc/serport.txt Signed-off-by: Calvin Lee <cyrus296@gmail.com> --- I included a slight code-style change in this commit because it seemed close enough to the code I was editing to be relevant. If not, I can change the commit to not include this change. hw/char/serial.c | 23 +++++++++++++---------- 1 file changed, 13 insertions(+), 10 deletions(-) diff --git a/hw/char/serial.c b/hw/char/serial.c index 2c080c9862..4159a46a2f 100644 --- a/hw/char/serial.c +++ b/hw/char/serial.c @@ -150,13 +150,10 @@ static void serial_update_irq(SerialState *s) static void serial_update_parameters(SerialState *s) { - int speed, parity, data_bits, stop_bits, frame_size; + float speed; + int parity, data_bits, stop_bits, frame_size; QEMUSerialSetParams ssp; - if (s->divider == 0 || s->divider > s->baudbase) { - return; - } - /* Start bit. */ frame_size = 1; if (s->lcr & 0x08) { @@ -169,14 +166,16 @@ static void serial_update_parameters(SerialState *s) } else { parity = 'N'; } - if (s->lcr & 0x04) + if (s->lcr & 0x04) { stop_bits = 2; - else + } else { stop_bits = 1; + } data_bits = (s->lcr & 0x03) + 5; frame_size += data_bits + stop_bits; - speed = s->baudbase / s->divider; + /* Zero divisor should give about 3500 baud */ + speed = (s->divider == 0) ? 3500 : (float) s->baudbase / s->divider; ssp.speed = speed; ssp.parity = parity; ssp.data_bits = data_bits; @@ -184,7 +183,7 @@ static void serial_update_parameters(SerialState *s) s->char_transmit_time = (NANOSECONDS_PER_SECOND / speed) * frame_size; qemu_chr_fe_ioctl(&s->chr, CHR_IOCTL_SERIAL_SET_PARAMS, &ssp); - DPRINTF("speed=%d parity=%c data=%d stop=%d\n", + DPRINTF("speed=%.2f parity=%c data=%d stop=%d\n", speed, parity, data_bits, stop_bits); } @@ -341,7 +340,11 @@ static void serial_ioport_write(void *opaque, hwaddr addr, uint64_t val, default: case 0: if (s->lcr & UART_LCR_DLAB) { - s->divider = (s->divider & 0xff00) | val; + if (size == 2) { + s->divider = (s->divider & 0xff00) | val; + } else if (size == 4) { + s->divider = val; + } serial_update_parameters(s); } else { s->thr = (uint8_t) val; -- 2.17.0 ^ permalink raw reply related [flat|nested] 6+ messages in thread
* [Qemu-devel] [PATCH RFC v2 2/2] PC Chipset: Send serial bytes at correct rate 2018-05-12 0:05 [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation Calvin Lee 2018-05-12 0:05 ` [Qemu-devel] [PATCH RFC v2 1/2] PC Chipset: Improve serial divisor calculation Calvin Lee @ 2018-05-12 0:05 ` Calvin Lee 2018-05-14 19:13 ` Dr. David Alan Gilbert 2018-07-15 15:57 ` [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation Paolo Bonzini 2 siblings, 1 reply; 6+ messages in thread From: Calvin Lee @ 2018-05-12 0:05 UTC (permalink / raw) To: qemu-devel; +Cc: mst, pbonzini, dgilbert, Calvin Lee This fixes bug in QEMU such that UART bytes would be sent immediatly after being put in the THR regardless of the UART frequency (and divisor). Now they will be sent at the appropriate rate. Signed-off-by: Calvin Lee <cyrus296@gmail.com> --- I am not sure about VM migration here. I want to move a struct field from one VMStateDescription to a new one. I did this by bumping the version number on the old VMStateDescription, and kept the field as `_TEST`. If this is incorrect please let me know. hw/char/serial.c | 51 +++++++++++++++++++++++++++++++++++----- include/hw/char/serial.h | 2 ++ 2 files changed, 47 insertions(+), 6 deletions(-) diff --git a/hw/char/serial.c b/hw/char/serial.c index 4159a46a2f..542d949ccd 100644 --- a/hw/char/serial.c +++ b/hw/char/serial.c @@ -359,9 +359,8 @@ static void serial_ioport_write(void *opaque, hwaddr addr, uint64_t val, s->lsr &= ~UART_LSR_THRE; s->lsr &= ~UART_LSR_TEMT; serial_update_irq(s); - if (s->tsr_retry == 0) { - serial_xmit(s); - } + timer_mod(s->xmit_timeout_timer, + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + s->char_transmit_time); } break; case 1: @@ -586,6 +585,15 @@ static void serial_receive_break(SerialState *s) serial_update_irq(s); } +/* There is data to be sent in xmit_fifo or the thr */ +static void xmit_timeout_int(void *opaque) +{ + SerialState *s = opaque; + if (s->tsr_retry == 0) { + serial_xmit(s); + } +} + /* There's data in recv_fifo and s->rbr has not been read for 4 char transmit times */ static void fifo_timeout_int (void *opaque) { SerialState *s = opaque; @@ -723,15 +731,20 @@ static bool serial_tsr_needed(void *opaque) SerialState *s = (SerialState *)opaque; return s->tsr_retry != 0; } +static bool serial_tsr_thr_exists(void *opaque, int version_id) +{ + return version_id < 2; +} static const VMStateDescription vmstate_serial_tsr = { .name = "serial/tsr", - .version_id = 1, + .version_id = 2, .minimum_version_id = 1, .needed = serial_tsr_needed, .fields = (VMStateField[]) { VMSTATE_UINT32(tsr_retry, SerialState), - VMSTATE_UINT8(thr, SerialState), + /* Moved to `xmit_timeout_timer` */ + VMSTATE_UINT8_TEST(thr, SerialState, serial_tsr_thr_exists), VMSTATE_UINT8(tsr, SerialState), VMSTATE_END_OF_LIST() } @@ -772,6 +785,24 @@ static const VMStateDescription vmstate_serial_xmit_fifo = { } }; +static bool serial_xmit_timeout_timer_needed(void *opaque) +{ + SerialState *s = (SerialState *)opaque; + return timer_pending(s->xmit_timeout_timer); +} + +static const VMStateDescription vmstate_serial_xmit_timeout_timer = { + .name = "serial/xmit_timeout_timer", + .version_id = 1, + .minimum_version_id = 1, + .needed = serial_xmit_timeout_timer_needed, + .fields = (VMStateField[]) { + VMSTATE_UINT8(thr, SerialState), + VMSTATE_TIMER_PTR(xmit_timeout_timer, SerialState), + VMSTATE_END_OF_LIST() + } +}; + static bool serial_fifo_timeout_timer_needed(void *opaque) { SerialState *s = (SerialState *)opaque; @@ -849,6 +880,7 @@ const VMStateDescription vmstate_serial = { &vmstate_serial_tsr, &vmstate_serial_recv_fifo, &vmstate_serial_xmit_fifo, + &vmstate_serial_xmit_timeout_timer, &vmstate_serial_fifo_timeout_timer, &vmstate_serial_timeout_ipending, &vmstate_serial_poll, @@ -880,6 +912,7 @@ static void serial_reset(void *opaque) s->poll_msl = 0; s->timeout_ipending = 0; + timer_del(s->xmit_timeout_timer); timer_del(s->fifo_timeout_timer); timer_del(s->modem_status_poll); @@ -928,7 +961,10 @@ void serial_realize_core(SerialState *s, Error **errp) { s->modem_status_poll = timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *) serial_update_msl, s); - s->fifo_timeout_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *) fifo_timeout_int, s); + s->xmit_timeout_timer = + timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *) xmit_timeout_int, s); + s->fifo_timeout_timer = + timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *) fifo_timeout_int, s); qemu_register_reset(serial_reset, s); qemu_chr_fe_set_handlers(&s->chr, serial_can_receive1, serial_receive1, @@ -945,6 +981,9 @@ void serial_exit_core(SerialState *s) timer_del(s->modem_status_poll); timer_free(s->modem_status_poll); + timer_del(s->xmit_timeout_timer); + timer_free(s->xmit_timeout_timer); + timer_del(s->fifo_timeout_timer); timer_free(s->fifo_timeout_timer); diff --git a/include/hw/char/serial.h b/include/hw/char/serial.h index 0acfbbc382..09aece90fb 100644 --- a/include/hw/char/serial.h +++ b/include/hw/char/serial.h @@ -65,6 +65,8 @@ struct SerialState { /* Time when the last byte was successfully sent out of the tsr */ uint64_t last_xmit_ts; Fifo8 recv_fifo; + /* Time to read the next byte from the thr */ + QEMUTimer *xmit_timeout_timer; Fifo8 xmit_fifo; /* Interrupt trigger level for recv_fifo */ uint8_t recv_fifo_itl; -- 2.17.0 ^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [Qemu-devel] [PATCH RFC v2 2/2] PC Chipset: Send serial bytes at correct rate 2018-05-12 0:05 ` [Qemu-devel] [PATCH RFC v2 2/2] PC Chipset: Send serial bytes at correct rate Calvin Lee @ 2018-05-14 19:13 ` Dr. David Alan Gilbert 2018-05-14 19:39 ` Calvin Lee 0 siblings, 1 reply; 6+ messages in thread From: Dr. David Alan Gilbert @ 2018-05-14 19:13 UTC (permalink / raw) To: Calvin Lee; +Cc: qemu-devel, mst, pbonzini * Calvin Lee (cyrus296@gmail.com) wrote: > This fixes bug in QEMU such that UART bytes would be sent immediatly > after being put in the THR regardless of the UART frequency (and divisor). > Now they will be sent at the appropriate rate. > > Signed-off-by: Calvin Lee <cyrus296@gmail.com> Hi Calvin, I'll leave the details of the serial to Paolo, but for the migration some comments below. > --- > I am not sure about VM migration here. I want to move a struct field > from one VMStateDescription to a new one. I did this by bumping the > version number on the old VMStateDescription, and kept the field as > `_TEST`. If this is incorrect please let me know. > > hw/char/serial.c | 51 +++++++++++++++++++++++++++++++++++----- > include/hw/char/serial.h | 2 ++ > 2 files changed, 47 insertions(+), 6 deletions(-) > > diff --git a/hw/char/serial.c b/hw/char/serial.c > index 4159a46a2f..542d949ccd 100644 > --- a/hw/char/serial.c > +++ b/hw/char/serial.c > @@ -359,9 +359,8 @@ static void serial_ioport_write(void *opaque, hwaddr addr, uint64_t val, > s->lsr &= ~UART_LSR_THRE; > s->lsr &= ~UART_LSR_TEMT; > serial_update_irq(s); > - if (s->tsr_retry == 0) { > - serial_xmit(s); > - } > + timer_mod(s->xmit_timeout_timer, > + qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL) + s->char_transmit_time); > } > break; > case 1: > @@ -586,6 +585,15 @@ static void serial_receive_break(SerialState *s) > serial_update_irq(s); > } > > +/* There is data to be sent in xmit_fifo or the thr */ > +static void xmit_timeout_int(void *opaque) > +{ > + SerialState *s = opaque; > + if (s->tsr_retry == 0) { > + serial_xmit(s); > + } > +} > + > /* There's data in recv_fifo and s->rbr has not been read for 4 char transmit times */ > static void fifo_timeout_int (void *opaque) { > SerialState *s = opaque; > @@ -723,15 +731,20 @@ static bool serial_tsr_needed(void *opaque) > SerialState *s = (SerialState *)opaque; > return s->tsr_retry != 0; > } > +static bool serial_tsr_thr_exists(void *opaque, int version_id) > +{ > + return version_id < 2; > +} > > static const VMStateDescription vmstate_serial_tsr = { > .name = "serial/tsr", > - .version_id = 1, > + .version_id = 2, > .minimum_version_id = 1, > .needed = serial_tsr_needed, > .fields = (VMStateField[]) { > VMSTATE_UINT32(tsr_retry, SerialState), > - VMSTATE_UINT8(thr, SerialState), > + /* Moved to `xmit_timeout_timer` */ > + VMSTATE_UINT8_TEST(thr, SerialState, serial_tsr_thr_exists), So the question is, why move it? If I understand what you've got, then it's the same flag with the same semantics - but moving it you break migration compatibility - so unless you need to, just leave it where it is. I think it's probably better if you leave the version_id = 1, and actually keep serial_tsr_thr_exists just for compatibility. You just need to fill in a sane value so that if you migrate to an older version it doesn't confuse the old one. > VMSTATE_UINT8(tsr, SerialState), > VMSTATE_END_OF_LIST() > } > @@ -772,6 +785,24 @@ static const VMStateDescription vmstate_serial_xmit_fifo = { > } > }; > > +static bool serial_xmit_timeout_timer_needed(void *opaque) > +{ > + SerialState *s = (SerialState *)opaque; > + return timer_pending(s->xmit_timeout_timer); > +} > + If you add a property/flag on the device, and check the property in that _needed function, then we can make it so that we don't save that section for older machine types. Dave > +static const VMStateDescription vmstate_serial_xmit_timeout_timer = { > + .name = "serial/xmit_timeout_timer", > + .version_id = 1, > + .minimum_version_id = 1, > + .needed = serial_xmit_timeout_timer_needed, > + .fields = (VMStateField[]) { > + VMSTATE_UINT8(thr, SerialState), > + VMSTATE_TIMER_PTR(xmit_timeout_timer, SerialState), > + VMSTATE_END_OF_LIST() > + } > +}; > + > static bool serial_fifo_timeout_timer_needed(void *opaque) > { > SerialState *s = (SerialState *)opaque; > @@ -849,6 +880,7 @@ const VMStateDescription vmstate_serial = { > &vmstate_serial_tsr, > &vmstate_serial_recv_fifo, > &vmstate_serial_xmit_fifo, > + &vmstate_serial_xmit_timeout_timer, > &vmstate_serial_fifo_timeout_timer, > &vmstate_serial_timeout_ipending, > &vmstate_serial_poll, > @@ -880,6 +912,7 @@ static void serial_reset(void *opaque) > s->poll_msl = 0; > > s->timeout_ipending = 0; > + timer_del(s->xmit_timeout_timer); > timer_del(s->fifo_timeout_timer); > timer_del(s->modem_status_poll); > > @@ -928,7 +961,10 @@ void serial_realize_core(SerialState *s, Error **errp) > { > s->modem_status_poll = timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *) serial_update_msl, s); > > - s->fifo_timeout_timer = timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *) fifo_timeout_int, s); > + s->xmit_timeout_timer = > + timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *) xmit_timeout_int, s); > + s->fifo_timeout_timer = > + timer_new_ns(QEMU_CLOCK_VIRTUAL, (QEMUTimerCB *) fifo_timeout_int, s); > qemu_register_reset(serial_reset, s); > > qemu_chr_fe_set_handlers(&s->chr, serial_can_receive1, serial_receive1, > @@ -945,6 +981,9 @@ void serial_exit_core(SerialState *s) > timer_del(s->modem_status_poll); > timer_free(s->modem_status_poll); > > + timer_del(s->xmit_timeout_timer); > + timer_free(s->xmit_timeout_timer); > + > timer_del(s->fifo_timeout_timer); > timer_free(s->fifo_timeout_timer); > > diff --git a/include/hw/char/serial.h b/include/hw/char/serial.h > index 0acfbbc382..09aece90fb 100644 > --- a/include/hw/char/serial.h > +++ b/include/hw/char/serial.h > @@ -65,6 +65,8 @@ struct SerialState { > /* Time when the last byte was successfully sent out of the tsr */ > uint64_t last_xmit_ts; > Fifo8 recv_fifo; > + /* Time to read the next byte from the thr */ > + QEMUTimer *xmit_timeout_timer; > Fifo8 xmit_fifo; > /* Interrupt trigger level for recv_fifo */ > uint8_t recv_fifo_itl; > -- > 2.17.0 > -- Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Qemu-devel] [PATCH RFC v2 2/2] PC Chipset: Send serial bytes at correct rate 2018-05-14 19:13 ` Dr. David Alan Gilbert @ 2018-05-14 19:39 ` Calvin Lee 0 siblings, 0 replies; 6+ messages in thread From: Calvin Lee @ 2018-05-14 19:39 UTC (permalink / raw) To: Dr. David Alan Gilbert; +Cc: qemu-devel, mst, pbonzini > So the question is, why move it? If I understand what you've got, then > it's the same flag with the same semantics - but moving it you break > migration compatibility - so unless you need to, just leave it where it > is. Sorry if I did not make this clear. This commit will delay flushing the THR until after the new timer has expired. Before, it would be immediatly flushed into the TSR, and was saved by `vmstate_serial_tsr`. However, now if the `xmit_timeout_timer` is pending it needs to be saved, that's why I moved it into the new VMStateDescription. >If you add a property/flag on the device, and check the property in that >_needed function, then we can make it so that we don't save that section >for older machine types. I don't really understand. How do I make sure not to save this for older machines? Thanks, Calvin ^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation 2018-05-12 0:05 [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation Calvin Lee 2018-05-12 0:05 ` [Qemu-devel] [PATCH RFC v2 1/2] PC Chipset: Improve serial divisor calculation Calvin Lee 2018-05-12 0:05 ` [Qemu-devel] [PATCH RFC v2 2/2] PC Chipset: Send serial bytes at correct rate Calvin Lee @ 2018-07-15 15:57 ` Paolo Bonzini 2 siblings, 0 replies; 6+ messages in thread From: Paolo Bonzini @ 2018-07-15 15:57 UTC (permalink / raw) To: Calvin Lee, qemu-devel; +Cc: mst, dgilbert On 12/05/2018 02:05, Calvin Lee wrote: > While developing a serial implementation for my OS, I found several bugs > in QEMU's serial device. I confirmed (by testing on my x64 laptop) that > there are several inconsistancies between QEMU and hardware in this > regard. For both patches, I used "http://www.sci.muni.cz/docs/pc/serport.txt" > for reference. > > First, QEMU has several errors for setting the UART divisor that are > fixed in my first patch. Second, (and more importantly) QEMU does not > transmit serial bytes at the correct rate, and this is fixed in my > second patch. Hi Calvin, I am queuing the first patch. Regarding the second, it's been debated quite a few times over the years whether QEMU should respect the transmit rate or not. The main advantage of not doing so is that some OSes use the serial port pretty heavily (e.g. Linux for boot messages, or Windows for debugging), and throttling the serial port to just e.g. 115200 bps makes them unnecessarily slow. Paolo ^ permalink raw reply [flat|nested] 6+ messages in thread
end of thread, other threads:[~2018-07-15 15:57 UTC | newest] Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2018-05-12 0:05 [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation Calvin Lee 2018-05-12 0:05 ` [Qemu-devel] [PATCH RFC v2 1/2] PC Chipset: Improve serial divisor calculation Calvin Lee 2018-05-12 0:05 ` [Qemu-devel] [PATCH RFC v2 2/2] PC Chipset: Send serial bytes at correct rate Calvin Lee 2018-05-14 19:13 ` Dr. David Alan Gilbert 2018-05-14 19:39 ` Calvin Lee 2018-07-15 15:57 ` [Qemu-devel] [PATCH RFC v2 0/2] Fix UART serial implementation Paolo Bonzini
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.