| Message ID | 1397205713-14076-1-git-send-email-shaveta@freescale.com |
|---|---|
| State | Superseded |
| Delegated to: | Heiko Schocher |
| Headers | show |
Hello Shaveta Leekha, Am 11.04.2014 10:41, schrieb Shaveta Leekha: > Most of the I2C slaves support accesses in the typical style > that is : read/write series of bytes at particular address offset. > These transactions look like:" > (1) START:Address:Tx:Offset:RESTART:Address[0..4]:Tx/Rx:data[0..n]:STOP" > > However there are certain devices which support accesses in > terms of the transactions as follows: > (2) "START:Address:Tx:Txdata[0..n1]:Clock_stretching: > RESTART:Address:Rx:data[0..n2]" > Here Txdata is typically a command and some associated data, > similarly Rxdata could be command status plus some data received > as a response to the command sent. > > Type (1) transactions are currently supportd in the > i2c driver using i2c_read and i2c_write APIs. I2C EEPROMs, > RTC, etc fall in this category. > > To handle type (2) along with type (1) transactions, > i2c_read() function has been modified. > > Signed-off-by: Shaveta Leekha<shaveta@freescale.com> > Signed-off-by: Poonam Aggrwal<poonam.aggrwal@freescale.com> > --- > drivers/i2c/fsl_i2c.c | 41 ++++++++++++++++++++++++++++++++++------- > 1 files changed, 34 insertions(+), 7 deletions(-) Hmm? Is this v2 of the following patch?: [U-Boot] 2c: modify i2c_read API to handle multi-bytes writes http://patchwork.ozlabs.org/patch/336911/ If so, please stick to the follow rules, for sending updated patches: http://www.denx.de/wiki/view/U-Boot/Patches#Sending_updated_patch_versions Thanks! bye, Heiko
Heiko Schocher <hs <at> denx.de> writes: > > Hmm? Is this v2 of the following patch?: > > [U-Boot] 2c: modify i2c_read API to handle multi-bytes writes > http://patchwork.ozlabs.org/patch/336911/ > > If so, please stick to the follow rules, for sending updated > patches: > > http://www.denx.de/wiki/view/U-Boot/Patches#Sending_updated_patch_versions > > Thanks! > > bye, > Heiko Yes, this the the v2 of this patch. I missed adding the reference here. I have rejected the older patch now, as the subject line was also little different. Should I send the v2/v3 of this new patch " [U-Boot] [PATCH] drivers/i2c/fsl_i2c: modify i2c_read to handle multi-byte write" now with the older references and changes mentioned? Or this patch would work? Regards, Shaveta
diff --git a/drivers/i2c/fsl_i2c.c b/drivers/i2c/fsl_i2c.c index 291ad94..aa159f8 100644 --- a/drivers/i2c/fsl_i2c.c +++ b/drivers/i2c/fsl_i2c.c @@ -423,18 +423,45 @@ fsl_i2c_read(struct i2c_adapter *adap, u8 dev, uint addr, int alen, u8 *data, struct fsl_i2c *device = (struct fsl_i2c *)i2c_dev[adap->hwadapnr]; int i = -1; /* signal error */ u8 *a = (u8*)&addr; + int len = alen * -1; if (i2c_wait4bus(adap) < 0) return -1; - if ((!length || alen > 0) - && i2c_write_addr(adap, dev, I2C_WRITE_BIT, 0) != 0 - && __i2c_write(adap, &a[4 - alen], alen) == alen) - i = 0; /* No error so far */ + /* To handle the need of I2C devices that require to write few bytes + * (more than 4 bytes of address as in the case of else part) + * of data before reading, Negative equivalent of length(bytes to write) + * is passed, but used the +ve part of len for writing data + */ + if (alen < 0) { + /* Generate a START and send the Address and + * the Tx Bytes to the slave. + * "START: Address: Write bytes data[len]" + * IF part supports writing any number of bytes in contrast + * to the else part, which supports writing address offset + * of upto 4 bytes only. + * bytes that need to be written are passed in + * "data", which will eventually keep the data READ, + * after writing the len bytes out of it + */ + if (i2c_write_addr(adap, dev, I2C_WRITE_BIT, 0) != 0) + i = __i2c_write(adap, data, len); + + if (i != len) + return -1; - if (length && - i2c_write_addr(adap, dev, I2C_READ_BIT, alen ? 1 : 0) != 0) - i = __i2c_read(adap, data, length); + if (length && i2c_write_addr(adap, dev, I2C_READ_BIT, 1) != 0) + i = __i2c_read(adap, data, length); + } else { + if ((!length || alen > 0) && + i2c_write_addr(adap, dev, I2C_WRITE_BIT, 0) != 0 && + __i2c_write(adap, &a[4 - alen], alen) == alen) + i = 0; /* No error so far */ + + if (length && + i2c_write_addr(adap, dev, I2C_READ_BIT, alen ? 1 : 0) != 0) + i = __i2c_read(adap, data, length); + } writeb(I2C_CR_MEN, &device->cr);
Most of the I2C slaves support accesses in the typical style that is : read/write series of bytes at particular address offset. These transactions look like:" (1) START:Address:Tx:Offset:RESTART:Address[0..4]:Tx/Rx:data[0..n]:STOP" However there are certain devices which support accesses in terms of the transactions as follows: (2) "START:Address:Tx:Txdata[0..n1]:Clock_stretching: RESTART:Address:Rx:data[0..n2]" Here Txdata is typically a command and some associated data, similarly Rxdata could be command status plus some data received as a response to the command sent. Type (1) transactions are currently supportd in the i2c driver using i2c_read and i2c_write APIs. I2C EEPROMs, RTC, etc fall in this category. To handle type (2) along with type (1) transactions, i2c_read() function has been modified. Signed-off-by: Shaveta Leekha <shaveta@freescale.com> Signed-off-by: Poonam Aggrwal <poonam.aggrwal@freescale.com> --- drivers/i2c/fsl_i2c.c | 41 ++++++++++++++++++++++++++++++++++------- 1 files changed, 34 insertions(+), 7 deletions(-)