| Message ID | 1658508510-15400-2-git-send-email-zhouyanjie@wanyeetech.com (mailing list archive) |
|---|---|
| State | Handled Elsewhere |
| Headers | show |
| Series | Add SFC support for Ingenic SoCs. | expand |
Hello! On 7/22/22 7:48 PM, 周琰杰 (Zhou Yanjie) wrote: > With advanced controllers (such as Ingenic SFC), it is possible to poll > the status register of the device. This could be done to offload the CPU > during a erase or write operation. Make use of spi-mem poll status APIs > to handle this feature. > > Previously, when erasing large area (e.g. 32MiB), in non-offload case, > CPU load could reach ~90% and would generate ~3.92 million interrupts, > now it decrease to ~15% CPU load and 0.15 million interrupts. > > This should also fix the high CPU usage for system which don't have a > dedicated poll status block logic (decrease to ~80% CPU load and ~1.61 > million interrupts.). > > Signed-off-by: 周琰杰 (Zhou Yanjie) <zhouyanjie@wanyeetech.com> > --- > drivers/mtd/spi-nor/core.c | 42 ++++++++++++++++++++++++++++++++---------- > 1 file changed, 32 insertions(+), 10 deletions(-) > > diff --git a/drivers/mtd/spi-nor/core.c b/drivers/mtd/spi-nor/core.c > index 502967c..6a31132 100644 > --- a/drivers/mtd/spi-nor/core.c > +++ b/drivers/mtd/spi-nor/core.c > @@ -617,19 +617,41 @@ static int spi_nor_wait_till_ready_with_timeout(struct spi_nor *nor, > unsigned long deadline; > int timeout = 0, ret; > > - deadline = jiffies + timeout_jiffies; > + if (nor->spimem && !nor->params->ready) { > + struct spi_mem_op op = SPI_MEM_OP(SPI_MEM_OP_CMD(SPINOR_OP_RDSR, 0), > + SPI_MEM_OP_NO_ADDR, > + SPI_MEM_OP_NO_DUMMY, > + SPI_MEM_OP_DATA_IN(1, nor->bouncebuf, 0)); Strange indentation... [...] > + return spi_mem_poll_status(nor->spimem, &op, SR_WIP, 0, 0, 10, > + jiffies_to_msecs(timeout_jiffies)); Here as well... [...] MBR, Sergey
diff --git a/drivers/mtd/spi-nor/core.c b/drivers/mtd/spi-nor/core.c index 502967c..6a31132 100644 --- a/drivers/mtd/spi-nor/core.c +++ b/drivers/mtd/spi-nor/core.c @@ -617,19 +617,41 @@ static int spi_nor_wait_till_ready_with_timeout(struct spi_nor *nor, unsigned long deadline; int timeout = 0, ret; - deadline = jiffies + timeout_jiffies; + if (nor->spimem && !nor->params->ready) { + struct spi_mem_op op = SPI_MEM_OP(SPI_MEM_OP_CMD(SPINOR_OP_RDSR, 0), + SPI_MEM_OP_NO_ADDR, + SPI_MEM_OP_NO_DUMMY, + SPI_MEM_OP_DATA_IN(1, nor->bouncebuf, 0)); - while (!timeout) { - if (time_after_eq(jiffies, deadline)) - timeout = 1; + if (nor->reg_proto == SNOR_PROTO_8_8_8_DTR) { + op.addr.nbytes = nor->params->rdsr_addr_nbytes; + op.dummy.nbytes = nor->params->rdsr_dummy; + /* + * We don't want to read only one byte in DTR mode. So, + * read 2 and then discard the second byte. + */ + op.data.nbytes = 2; + } - ret = spi_nor_ready(nor); - if (ret < 0) - return ret; - if (ret) - return 0; + spi_nor_spimem_setup_op(nor, &op, nor->reg_proto); + + return spi_mem_poll_status(nor->spimem, &op, SR_WIP, 0, 0, 10, + jiffies_to_msecs(timeout_jiffies)); + } else { + deadline = jiffies + timeout_jiffies; - cond_resched(); + while (!timeout) { + if (time_after_eq(jiffies, deadline)) + timeout = 1; + + ret = spi_nor_ready(nor); + if (ret < 0) + return ret; + if (ret) + return 0; + + cond_resched(); + } } dev_dbg(nor->dev, "flash operation timed out\n");
With advanced controllers (such as Ingenic SFC), it is possible to poll the status register of the device. This could be done to offload the CPU during a erase or write operation. Make use of spi-mem poll status APIs to handle this feature. Previously, when erasing large area (e.g. 32MiB), in non-offload case, CPU load could reach ~90% and would generate ~3.92 million interrupts, now it decrease to ~15% CPU load and 0.15 million interrupts. This should also fix the high CPU usage for system which don't have a dedicated poll status block logic (decrease to ~80% CPU load and ~1.61 million interrupts.). Signed-off-by: 周琰杰 (Zhou Yanjie) <zhouyanjie@wanyeetech.com> --- drivers/mtd/spi-nor/core.c | 42 ++++++++++++++++++++++++++++++++---------- 1 file changed, 32 insertions(+), 10 deletions(-)