| Message ID | 20190415220218.10551-2-tpiepho@impinj.com |
|---|---|
| State | Changes Requested |
| Delegated to: | Heiko Schocher |
| Headers | show |
| Series | [U-Boot,1/2] i2c: mxc_i2c: Document how non-DM functions work | expand |
Hello Trent, Am 16.04.2019 um 00:02 schrieb Trent Piepho: > This is an old driver that supports both device mapped and non-mapped > mode, and covers a wide range of hardware. It's hard to change without > risking breaking something. I have to tried to be exceedingly detailed > in this patch, so please excuse the length of the commit essay that > follows. > > In device mapped mode the I2C xfer function does not handle plain read, > and some other, transfers correctly. > > What it can't handle are transactions that: > Start with a read, or, > Have a write followed by a read, or, > Have more than one read in a row. > > The common I2C/SMBUS read register and write register transactions > always start with a write, followed by a write or a read, and then end. > These work, so the bug is not apparent for most I2C slaves that only use > these common xfer forms. > > The existing xfer loop initializes by sending the chip address in write > mode after it deals with bus arbitration and master setup. When > processing each message, if the next message will be a read, it sends a > repeated start followed by the chip address in read mode after the > current message. > > Obviously, this does not work if the first message is a read, as the > chip is always addressed in write mode initially by i2c_init_transfer(). > > A write following a read does not work because the repeated start is > only sent when the next message is a read. There is no logic to send it > when the current message is a read and next is write. It should be sent > every time the bus changes direction. > > The ability to use a plain read was added to this driver in > commit 2feec4eafd40 ("imx: mxc_i2c: tweak the i2c transfer method"), > but this applied only the non-DM code path. > > This patch fixes the DM code path. The xfer function will call > i2c_init_transfer() with an alen of -1 to avoid sending the chip > address. The same way the non-DM code achieves this. The xfer > function's message loop will send the address and mode before each > message if the bus changes direction, and on the first message. > > When reading data, the master hardware is one byte ahead of what we > receive. I.e., reading a byte from the data register returns a byte > *already received* by the master, and causes the master to start the RX > of the *next* byte. Therefor, before we read the final byte of a > message, we must tell the master what to do next. I add a "last" flag > to i2c_read_data() to tell it if the message is to be followed by a stop > or a repeated start. When last == true it acts exactly as before. > > The non-DM code can only create an xfer where the read, if any, is the > final message of the xfer. And so the only callsite of i2c_read_data() > in the non-DM code has the "last" parameter as true. Therefore, this > change has no effect on the non-DM code. As all other changes are in > the DM xfer function, which is not even compiled in non-DM code, I am > confident that this patch has no effect on boards not using I2C_DM. > This greatly reduces the change of hardware that could be affected. > > For DM boards, I have verified every transaction the "i2c" command can > create on a scope and they are all exactly as they are supposed to be. > I also tested write->read->write, which isn't possible with the i2c > command, and it works as well. I didn't fix multiple reads in a row, as > it's a lot more invasive and obviously no one has every wanted them > since they've never worked. It didn't seem like the extra complexity > was justified to support something no one uses. > > Cc: Nandor Han <nandor.han@ge.com> > Cc: Heiko Schocher <hs@denx.de> > Cc: Stefano Babic <sbabic@denx.de> > Cc: Fabio Estevam <festevam@gmail.com> > Cc: Breno Matheus Lima <brenomatheus@gmail.com> > Signed-off-by: Trent Piepho <tpiepho@impinj.com> > --- > drivers/i2c/mxc_i2c.c | 95 ++++++++++++++++++++++++++++++++++----------------- > 1 file changed, 64 insertions(+), 31 deletions(-) Thanks for this work! Your patch has total: 64 errors, 2 warnings, 0 checks, 145 lines checked Please fix and send a v2. It would be good to become here some Tested by tags ... Reviewed-by: Heiko Schocher <hs@denx.de> bye, Heiko > diff --git a/drivers/i2c/mxc_i2c.c b/drivers/i2c/mxc_i2c.c > index f17a47f600..23119cce65 100644 > --- a/drivers/i2c/mxc_i2c.c > +++ b/drivers/i2c/mxc_i2c.c > @@ -482,8 +482,13 @@ static int i2c_write_data(struct mxc_i2c_bus *i2c_bus, u8 chip, const u8 *buf, > return ret; > } > > +/* Will generate a STOP after the last byte if "last" is true, i.e. this is the > + * final message of a transaction. If not, it switches the bus back to TX mode > + * and does not send a STOP, leaving the bus in a state where a repeated start > + * and address can be sent for another message. > + */ > static int i2c_read_data(struct mxc_i2c_bus *i2c_bus, uchar chip, uchar *buf, > - int len) > + int len, bool last) > { > int ret; > unsigned int temp; > @@ -513,17 +518,31 @@ static int i2c_read_data(struct mxc_i2c_bus *i2c_bus, uchar chip, uchar *buf, > return ret; > } > > - /* > - * It must generate STOP before read I2DR to prevent > - * controller from generating another clock cycle > - */ > if (i == (len - 1)) { > - i2c_imx_stop(i2c_bus); > + /* Final byte has already been received by master! When > + * we read it from I2DR, the master will start another > + * cycle. We must program it first to send a STOP or > + * switch to TX to avoid this. > + */ > + if (last) { > + i2c_imx_stop(i2c_bus); > + } else { > + /* Final read, no stop, switch back to tx */ > + temp = readb(base + (I2CR << reg_shift)); > + temp |= I2CR_MTX | I2CR_TX_NO_AK; > + writeb(temp, base + (I2CR << reg_shift)); > + } > } else if (i == (len - 2)) { > + /* Master has already recevied penultimate byte. When > + * we read it from I2DR, master will start RX of final > + * byte. We must set TX_NO_AK now so it does not ACK > + * that final byte. > + */ > temp = readb(base + (I2CR << reg_shift)); > temp |= I2CR_TX_NO_AK; > writeb(temp, base + (I2CR << reg_shift)); > } > + > writeb(I2SR_IIF_CLEAR, base + (I2SR << reg_shift)); > buf[i] = readb(base + (I2DR << reg_shift)); > } > @@ -533,7 +552,9 @@ static int i2c_read_data(struct mxc_i2c_bus *i2c_bus, uchar chip, uchar *buf, > debug(" 0x%02x", buf[ret]); > debug("\n"); > > - i2c_imx_stop(i2c_bus); > + /* It is not clear to me that this is necessary */ > + if (last) > + i2c_imx_stop(i2c_bus); > return 0; > } > > @@ -585,7 +606,7 @@ static int bus_i2c_read(struct mxc_i2c_bus *i2c_bus, u8 chip, u32 addr, > return ret; > } > > - ret = i2c_read_data(i2c_bus, chip, buf, len); > + ret = i2c_read_data(i2c_bus, chip, buf, len, true); > > i2c_imx_stop(i2c_bus); > return ret; > @@ -939,42 +960,54 @@ static int mxc_i2c_xfer(struct udevice *bus, struct i2c_msg *msg, int nmsgs) > ulong base = i2c_bus->base; > int reg_shift = i2c_bus->driver_data & I2C_QUIRK_FLAG ? > VF610_I2C_REGSHIFT : IMX_I2C_REGSHIFT; > + int read_mode; > > - /* > - * Here the 3rd parameter addr and the 4th one alen are set to 0, > - * because here we only want to send out chip address. The register > - * address is wrapped in msg. > + /* Here address len is set to -1 to not send any address at first. > + * Otherwise i2c_init_transfer will send the chip address with write > + * mode set. This is wrong if the 1st message is read. > */ > - ret = i2c_init_transfer(i2c_bus, msg->addr, 0, 0); > + ret = i2c_init_transfer(i2c_bus, msg->addr, 0, -1); > if (ret < 0) { > debug("i2c_init_transfer error: %d\n", ret); > return ret; > } > > + read_mode = -1; /* So it's always different on the first message */ > for (; nmsgs > 0; nmsgs--, msg++) { > - bool next_is_read = nmsgs > 1 && (msg[1].flags & I2C_M_RD); > - debug("i2c_xfer: chip=0x%x, len=0x%x\n", msg->addr, msg->len); > - if (msg->flags & I2C_M_RD) > - ret = i2c_read_data(i2c_bus, msg->addr, msg->buf, > - msg->len); > - else { > - ret = i2c_write_data(i2c_bus, msg->addr, msg->buf, > - msg->len); > - if (ret) > - break; > - if (next_is_read) { > - /* Reuse ret */ > + const int msg_is_read = !!(msg->flags & I2C_M_RD); > + > + debug("i2c_xfer: chip=0x%x, len=0x%x, dir=%c\n", msg->addr, > + msg->len, msg_is_read ? 'R' : 'W'); > + > + if (msg_is_read != read_mode) { > + /* Send repeated start if not 1st message */ > + if (read_mode != -1) { > + debug("i2c_xfer: [RSTART]\n"); > ret = readb(base + (I2CR << reg_shift)); > ret |= I2CR_RSTA; > writeb(ret, base + (I2CR << reg_shift)); > - > - ret = tx_byte(i2c_bus, (msg->addr << 1) | 1); > - if (ret < 0) { > - i2c_imx_stop(i2c_bus); > - break; > - } > } > + debug("i2c_xfer: [ADDR %02x | %c]\n", msg->addr, > + msg_is_read ? 'R' : 'W'); > + ret = tx_byte(i2c_bus, (msg->addr << 1) | msg_is_read); > + if (ret < 0) { > + debug("i2c_xfer: [STOP]\n"); > + i2c_imx_stop(i2c_bus); > + break; > + } > + read_mode = msg_is_read; > } > + > + if (msg->flags & I2C_M_RD) > + ret = i2c_read_data(i2c_bus, msg->addr, msg->buf, > + msg->len, nmsgs == 1 || > + (msg->flags & I2C_M_STOP)); > + else > + ret = i2c_write_data(i2c_bus, msg->addr, msg->buf, > + msg->len); > + > + if (ret < 0) > + break; > } > > if (ret) >
On Tue, 2019-04-30 at 06:34 +0200, Heiko Schocher wrote: > Hello Trent, > > Am 16.04.2019 um 00:02 schrieb Trent Piepho: > > This is an old driver that supports both device mapped and non-mapped > > mode, and covers a wide range of hardware. It's hard to change without > > risking breaking something. I have to tried to be exceedingly detailed > > in this patch, so please excuse the length of the commit essay that > > follows. > > > Thanks for this work! > > Your patch has > > total: 64 errors, 2 warnings, 0 checks, 145 lines checked > > Please fix and send a v2. > > It would be good to become here some Tested by tags ... > > Reviewed-by: Heiko Schocher <hs@denx.de> My patches are fine before I send them. They are getting DOS line endings added in transit. It seems Microsoft has recently changed their mail server used for office365 to automatically add DOS line endings to all text emails. It didn't use to do this. It seems the only way to avoid this is to use base64 encoding with git send-email. It works went I send a test message to myself and export from Evolution. No way to test if the list software can handle it without re-sending. The patches should be automatically fixed by git am, as it will strip DOS endings by default.
Hello Trent, Am 30.04.2019 um 18:04 schrieb Trent Piepho: > On Tue, 2019-04-30 at 06:34 +0200, Heiko Schocher wrote: >> Hello Trent, >> >> Am 16.04.2019 um 00:02 schrieb Trent Piepho: >>> This is an old driver that supports both device mapped and non-mapped >>> mode, and covers a wide range of hardware. It's hard to change without >>> risking breaking something. I have to tried to be exceedingly detailed >>> in this patch, so please excuse the length of the commit essay that >>> follows. >>> >> Thanks for this work! >> >> Your patch has >> >> total: 64 errors, 2 warnings, 0 checks, 145 lines checked >> >> Please fix and send a v2. >> >> It would be good to become here some Tested by tags ... >> >> Reviewed-by: Heiko Schocher <hs@denx.de> > > My patches are fine before I send them. They are getting DOS line > endings added in transit. It seems Microsoft has recently changed > their mail server used for office365 to automatically add DOS line > endings to all text emails. It didn't use to do this. It seems the > only way to avoid this is to use base64 encoding with git send-email. > It works went I send a test message to myself and export from > Evolution. No way to test if the list software can handle it without > re-sending. > > The patches should be automatically fixed by git am, as it will strip > DOS endings by default. Unfortunately, your v2 version has the same problem :-( I pushed them now to: https://github.com/hsdenx/u-boot-i2c/commits/master as I started a travis build, see: https://travis-ci.org/hsdenx/u-boot-i2c/builds/527155944 Can you please check, if your patches are OK? Thanks! bye, Heiko
On Thu, 2019-05-02 at 07:23 +0200, Heiko Schocher wrote: > Am 30.04.2019 um 18:04 schrieb Trent Piepho: > > On Tue, 2019-04-30 at 06:34 +0200, Heiko Schocher wrote: > > > Am 16.04.2019 um 00:02 schrieb Trent Piepho: > > > > This is an old driver that supports both device mapped and non-mapped > > > > mode, and covers a wide range of hardware. It's hard to change without > > > > risking breaking something. I have to tried to be exceedingly detailed > > > > in this patch, so please excuse the length of the commit essay that > > > > follows. > > > > > > > > > > > The patches should be automatically fixed by git am, as it will strip > > DOS endings by default. > > Unfortunately, your v2 version has the same problem :-( > > Can you please check, if your patches are OK? They look correct, thanks for applying them. I'm not sure where the mangling is happening. I assure you the patches I'm sending are correct. They arrive at both my work and gmail account unmangled. Yet somehow the patchwork mbox file has DOS line endings added to the diff portion but not the commit message. I wonder if the flaw is somewhere in in the denx list server or in patchwork? Of course, if no one else's patches get mangled, it's probably something about microsoft's email server that is the problem.
diff --git a/drivers/i2c/mxc_i2c.c b/drivers/i2c/mxc_i2c.c index f17a47f600..23119cce65 100644 --- a/drivers/i2c/mxc_i2c.c +++ b/drivers/i2c/mxc_i2c.c @@ -482,8 +482,13 @@ static int i2c_write_data(struct mxc_i2c_bus *i2c_bus, u8 chip, const u8 *buf, return ret; } +/* Will generate a STOP after the last byte if "last" is true, i.e. this is the + * final message of a transaction. If not, it switches the bus back to TX mode + * and does not send a STOP, leaving the bus in a state where a repeated start + * and address can be sent for another message. + */ static int i2c_read_data(struct mxc_i2c_bus *i2c_bus, uchar chip, uchar *buf, - int len) + int len, bool last) { int ret; unsigned int temp; @@ -513,17 +518,31 @@ static int i2c_read_data(struct mxc_i2c_bus *i2c_bus, uchar chip, uchar *buf, return ret; } - /* - * It must generate STOP before read I2DR to prevent - * controller from generating another clock cycle - */ if (i == (len - 1)) { - i2c_imx_stop(i2c_bus); + /* Final byte has already been received by master! When + * we read it from I2DR, the master will start another + * cycle. We must program it first to send a STOP or + * switch to TX to avoid this. + */ + if (last) { + i2c_imx_stop(i2c_bus); + } else { + /* Final read, no stop, switch back to tx */ + temp = readb(base + (I2CR << reg_shift)); + temp |= I2CR_MTX | I2CR_TX_NO_AK; + writeb(temp, base + (I2CR << reg_shift)); + } } else if (i == (len - 2)) { + /* Master has already recevied penultimate byte. When + * we read it from I2DR, master will start RX of final + * byte. We must set TX_NO_AK now so it does not ACK + * that final byte. + */ temp = readb(base + (I2CR << reg_shift)); temp |= I2CR_TX_NO_AK; writeb(temp, base + (I2CR << reg_shift)); } + writeb(I2SR_IIF_CLEAR, base + (I2SR << reg_shift)); buf[i] = readb(base + (I2DR << reg_shift)); } @@ -533,7 +552,9 @@ static int i2c_read_data(struct mxc_i2c_bus *i2c_bus, uchar chip, uchar *buf, debug(" 0x%02x", buf[ret]); debug("\n"); - i2c_imx_stop(i2c_bus); + /* It is not clear to me that this is necessary */ + if (last) + i2c_imx_stop(i2c_bus); return 0; } @@ -585,7 +606,7 @@ static int bus_i2c_read(struct mxc_i2c_bus *i2c_bus, u8 chip, u32 addr, return ret; } - ret = i2c_read_data(i2c_bus, chip, buf, len); + ret = i2c_read_data(i2c_bus, chip, buf, len, true); i2c_imx_stop(i2c_bus); return ret; @@ -939,42 +960,54 @@ static int mxc_i2c_xfer(struct udevice *bus, struct i2c_msg *msg, int nmsgs) ulong base = i2c_bus->base; int reg_shift = i2c_bus->driver_data & I2C_QUIRK_FLAG ? VF610_I2C_REGSHIFT : IMX_I2C_REGSHIFT; + int read_mode; - /* - * Here the 3rd parameter addr and the 4th one alen are set to 0, - * because here we only want to send out chip address. The register - * address is wrapped in msg. + /* Here address len is set to -1 to not send any address at first. + * Otherwise i2c_init_transfer will send the chip address with write + * mode set. This is wrong if the 1st message is read. */ - ret = i2c_init_transfer(i2c_bus, msg->addr, 0, 0); + ret = i2c_init_transfer(i2c_bus, msg->addr, 0, -1); if (ret < 0) { debug("i2c_init_transfer error: %d\n", ret); return ret; } + read_mode = -1; /* So it's always different on the first message */ for (; nmsgs > 0; nmsgs--, msg++) { - bool next_is_read = nmsgs > 1 && (msg[1].flags & I2C_M_RD); - debug("i2c_xfer: chip=0x%x, len=0x%x\n", msg->addr, msg->len); - if (msg->flags & I2C_M_RD) - ret = i2c_read_data(i2c_bus, msg->addr, msg->buf, - msg->len); - else { - ret = i2c_write_data(i2c_bus, msg->addr, msg->buf, - msg->len); - if (ret) - break; - if (next_is_read) { - /* Reuse ret */ + const int msg_is_read = !!(msg->flags & I2C_M_RD); + + debug("i2c_xfer: chip=0x%x, len=0x%x, dir=%c\n", msg->addr, + msg->len, msg_is_read ? 'R' : 'W'); + + if (msg_is_read != read_mode) { + /* Send repeated start if not 1st message */ + if (read_mode != -1) { + debug("i2c_xfer: [RSTART]\n"); ret = readb(base + (I2CR << reg_shift)); ret |= I2CR_RSTA; writeb(ret, base + (I2CR << reg_shift)); - - ret = tx_byte(i2c_bus, (msg->addr << 1) | 1); - if (ret < 0) { - i2c_imx_stop(i2c_bus); - break; - } } + debug("i2c_xfer: [ADDR %02x | %c]\n", msg->addr, + msg_is_read ? 'R' : 'W'); + ret = tx_byte(i2c_bus, (msg->addr << 1) | msg_is_read); + if (ret < 0) { + debug("i2c_xfer: [STOP]\n"); + i2c_imx_stop(i2c_bus); + break; + } + read_mode = msg_is_read; } + + if (msg->flags & I2C_M_RD) + ret = i2c_read_data(i2c_bus, msg->addr, msg->buf, + msg->len, nmsgs == 1 || + (msg->flags & I2C_M_STOP)); + else + ret = i2c_write_data(i2c_bus, msg->addr, msg->buf, + msg->len); + + if (ret < 0) + break; } if (ret)
This is an old driver that supports both device mapped and non-mapped mode, and covers a wide range of hardware. It's hard to change without risking breaking something. I have to tried to be exceedingly detailed in this patch, so please excuse the length of the commit essay that follows. In device mapped mode the I2C xfer function does not handle plain read, and some other, transfers correctly. What it can't handle are transactions that: Start with a read, or, Have a write followed by a read, or, Have more than one read in a row. The common I2C/SMBUS read register and write register transactions always start with a write, followed by a write or a read, and then end. These work, so the bug is not apparent for most I2C slaves that only use these common xfer forms. The existing xfer loop initializes by sending the chip address in write mode after it deals with bus arbitration and master setup. When processing each message, if the next message will be a read, it sends a repeated start followed by the chip address in read mode after the current message. Obviously, this does not work if the first message is a read, as the chip is always addressed in write mode initially by i2c_init_transfer(). A write following a read does not work because the repeated start is only sent when the next message is a read. There is no logic to send it when the current message is a read and next is write. It should be sent every time the bus changes direction. The ability to use a plain read was added to this driver in commit 2feec4eafd40 ("imx: mxc_i2c: tweak the i2c transfer method"), but this applied only the non-DM code path. This patch fixes the DM code path. The xfer function will call i2c_init_transfer() with an alen of -1 to avoid sending the chip address. The same way the non-DM code achieves this. The xfer function's message loop will send the address and mode before each message if the bus changes direction, and on the first message. When reading data, the master hardware is one byte ahead of what we receive. I.e., reading a byte from the data register returns a byte *already received* by the master, and causes the master to start the RX of the *next* byte. Therefor, before we read the final byte of a message, we must tell the master what to do next. I add a "last" flag to i2c_read_data() to tell it if the message is to be followed by a stop or a repeated start. When last == true it acts exactly as before. The non-DM code can only create an xfer where the read, if any, is the final message of the xfer. And so the only callsite of i2c_read_data() in the non-DM code has the "last" parameter as true. Therefore, this change has no effect on the non-DM code. As all other changes are in the DM xfer function, which is not even compiled in non-DM code, I am confident that this patch has no effect on boards not using I2C_DM. This greatly reduces the change of hardware that could be affected. For DM boards, I have verified every transaction the "i2c" command can create on a scope and they are all exactly as they are supposed to be. I also tested write->read->write, which isn't possible with the i2c command, and it works as well. I didn't fix multiple reads in a row, as it's a lot more invasive and obviously no one has every wanted them since they've never worked. It didn't seem like the extra complexity was justified to support something no one uses. Cc: Nandor Han <nandor.han@ge.com> Cc: Heiko Schocher <hs@denx.de> Cc: Stefano Babic <sbabic@denx.de> Cc: Fabio Estevam <festevam@gmail.com> Cc: Breno Matheus Lima <brenomatheus@gmail.com> Signed-off-by: Trent Piepho <tpiepho@impinj.com> --- drivers/i2c/mxc_i2c.c | 95 ++++++++++++++++++++++++++++++++++----------------- 1 file changed, 64 insertions(+), 31 deletions(-)