qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
* [Qemu-devel] [PATCH] i2c: Add asserts for second smbus i2c_start_transfer()
@ 2016-10-24 15:10 minyard
  2016-10-24 15:15 ` Peter Maydell
  0 siblings, 1 reply; 2+ messages in thread
From: minyard @ 2016-10-24 15:10 UTC (permalink / raw)
  To: Peter Maydell
  Cc: QEMU Developers, Corey Minyard, Alistair Francis, KONRAD Frederic

From: Corey Minyard <cminyard@mvista.com>

Some SMBus operations restart the transfer to convert from
write to read mode without an intervening i2c_end_transfer().
The second call cannot fail, so the return code is unchecked,
but this causes Coverity to complain.  So add some asserts
and documentation about this.

Signed-off-by: Corey Minyard <cminyard@mvista.com>
---
 hw/i2c/core.c  | 5 ++++-
 hw/i2c/smbus.c | 6 +++---
 2 files changed, 7 insertions(+), 4 deletions(-)

I tested this with the IPMI SMbus driver for a while.

diff --git a/hw/i2c/core.c b/hw/i2c/core.c
index bd8f167..b9dea79 100644
--- a/hw/i2c/core.c
+++ b/hw/i2c/core.c
@@ -88,7 +88,10 @@ int i2c_bus_busy(I2CBus *bus)
     return !QLIST_EMPTY(&bus->current_devs);
 }
 
-/* Returns non-zero if the address is not valid.  */
+/* Returns non-zero if the address is not valid.  If this is called
+   again without an intervening i2c_end_transfer(), like in the SMBus
+   case where the operation is switched from write to read, this
+   function will not rescan the bus and thus cannot fail. */
 /* TODO: Make this handle multiple masters.  */
 int i2c_start_transfer(I2CBus *bus, uint8_t address, int recv)
 {
diff --git a/hw/i2c/smbus.c b/hw/i2c/smbus.c
index 3979b3d..bf46d22 100644
--- a/hw/i2c/smbus.c
+++ b/hw/i2c/smbus.c
@@ -248,7 +248,7 @@ int smbus_read_byte(I2CBus *bus, uint8_t addr, uint8_t command)
         return -1;
     }
     i2c_send(bus, command);
-    i2c_start_transfer(bus, addr, 1);
+    assert(!i2c_start_transfer(bus, addr, 1));
     data = i2c_recv(bus);
     i2c_nack(bus);
     i2c_end_transfer(bus);
@@ -273,7 +273,7 @@ int smbus_read_word(I2CBus *bus, uint8_t addr, uint8_t command)
         return -1;
     }
     i2c_send(bus, command);
-    i2c_start_transfer(bus, addr, 1);
+    assert(!i2c_start_transfer(bus, addr, 1));
     data = i2c_recv(bus);
     data |= i2c_recv(bus) << 8;
     i2c_nack(bus);
@@ -302,7 +302,7 @@ int smbus_read_block(I2CBus *bus, uint8_t addr, uint8_t command, uint8_t *data)
         return -1;
     }
     i2c_send(bus, command);
-    i2c_start_transfer(bus, addr, 1);
+    assert(!i2c_start_transfer(bus, addr, 1));
     len = i2c_recv(bus);
     if (len > 32) {
         len = 0;
-- 
2.7.4

^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [Qemu-devel] [PATCH] i2c: Add asserts for second smbus i2c_start_transfer()
  2016-10-24 15:10 [Qemu-devel] [PATCH] i2c: Add asserts for second smbus i2c_start_transfer() minyard
@ 2016-10-24 15:15 ` Peter Maydell
  0 siblings, 0 replies; 2+ messages in thread
From: Peter Maydell @ 2016-10-24 15:15 UTC (permalink / raw)
  To: Corey Minyard
  Cc: QEMU Developers, Corey Minyard, Alistair Francis, KONRAD Frederic

On 24 October 2016 at 16:10,  <minyard@acm.org> wrote:
> From: Corey Minyard <cminyard@mvista.com>
>
> Some SMBus operations restart the transfer to convert from
> write to read mode without an intervening i2c_end_transfer().
> The second call cannot fail, so the return code is unchecked,
> but this causes Coverity to complain.  So add some asserts
> and documentation about this.
>
> Signed-off-by: Corey Minyard <cminyard@mvista.com>
> ---
>  hw/i2c/core.c  | 5 ++++-
>  hw/i2c/smbus.c | 6 +++---
>  2 files changed, 7 insertions(+), 4 deletions(-)
>
> I tested this with the IPMI SMbus driver for a while.
>
> diff --git a/hw/i2c/core.c b/hw/i2c/core.c
> index bd8f167..b9dea79 100644
> --- a/hw/i2c/core.c
> +++ b/hw/i2c/core.c
> @@ -88,7 +88,10 @@ int i2c_bus_busy(I2CBus *bus)
>      return !QLIST_EMPTY(&bus->current_devs);
>  }
>
> -/* Returns non-zero if the address is not valid.  */
> +/* Returns non-zero if the address is not valid.  If this is called
> +   again without an intervening i2c_end_transfer(), like in the SMBus
> +   case where the operation is switched from write to read, this
> +   function will not rescan the bus and thus cannot fail. */

A nit, but comment syntax convention in this file seems to be to have
this kind of long-comment:

/*
 * text text text
 * text text text
 */

>  /* TODO: Make this handle multiple masters.  */
>  int i2c_start_transfer(I2CBus *bus, uint8_t address, int recv)
>  {
> diff --git a/hw/i2c/smbus.c b/hw/i2c/smbus.c
> index 3979b3d..bf46d22 100644
> --- a/hw/i2c/smbus.c
> +++ b/hw/i2c/smbus.c
> @@ -248,7 +248,7 @@ int smbus_read_byte(I2CBus *bus, uint8_t addr, uint8_t command)
>          return -1;
>      }
>      i2c_send(bus, command);
> -    i2c_start_transfer(bus, addr, 1);
> +    assert(!i2c_start_transfer(bus, addr, 1));

Better not to put expressions that have side-effects inside
assert() -- if the assert() is compiled out then the
expression inside it will never be evaluated...

thanks
-- PMM

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2016-10-24 15:15 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2016-10-24 15:10 [Qemu-devel] [PATCH] i2c: Add asserts for second smbus i2c_start_transfer() minyard
2016-10-24 15:15 ` Peter Maydell

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).