Previous to OpenBSD 6.3 [1], fcntl(F_SETFL) is not permitted on memory
devices. Do not assert fcntl failures on OpenBSD.
This fixes:

  $ lm32-softmmu/qemu-system-lm32
  assertion "f != -1" failed: file "util/oslib-posix.c", line 247, function "qemu_set_nonblock"
  Abort trap (core dumped)

[1] The fix seems https://github.com/openbsd/src/commit/c2a35b387f9d3c
  "fcntl(F_SETFL) invokes the FIONBIO and FIOASYNC ioctls internally, so
  the memory devices (/dev/null, /dev/zero, etc) need to permit them."

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
---
 util/oslib-posix.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/util/oslib-posix.c b/util/oslib-posix.c
index 4ce1ba9ca4..064c3ae2f7 100644
--- a/util/oslib-posix.c
+++ b/util/oslib-posix.c
@@ -244,7 +244,9 @@ void qemu_set_nonblock(int fd)
     f = fcntl(fd, F_GETFL);
     assert(f != -1);
     f = fcntl(fd, F_SETFL, f | O_NONBLOCK);
+#ifndef __OpenBSD__
     assert(f != -1);
+#endif
 }
 
 int socket_set_fast_reuse(int fd)
-- 
2.20.1

Is the XXX in the subject meant to go on permanent record?

On 1/28/19 7:22 AM, Markus Armbruster wrote:
> Is the XXX in the subject meant to go on permanent record?

The original plan was to rename it as NOTFORMERGE before sending but I
forgot (late Friday). So the whole content of this patch isn't meant to
go on perm record ;)

Philippe Mathieu-Daudé <philmd@redhat.com> writes:

> Previous to OpenBSD 6.3 [1], fcntl(F_SETFL) is not permitted on memory
> devices. Do not assert fcntl failures on OpenBSD.
> This fixes:
>
>   $ lm32-softmmu/qemu-system-lm32
>   assertion "f != -1" failed: file "util/oslib-posix.c", line 247, function "qemu_set_nonblock"
>   Abort trap (core dumped)
>
> [1] The fix seems https://github.com/openbsd/src/commit/c2a35b387f9d3c
>   "fcntl(F_SETFL) invokes the FIONBIO and FIOASYNC ioctls internally, so
>   the memory devices (/dev/null, /dev/zero, etc) need to permit them."

I assume set_nonblock is called on more than just these special devices?
Is there anyway to check this on OpenBSD or is it just an anonymous fd
at this point?

>
> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
> ---
>  util/oslib-posix.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/util/oslib-posix.c b/util/oslib-posix.c
> index 4ce1ba9ca4..064c3ae2f7 100644
> --- a/util/oslib-posix.c
> +++ b/util/oslib-posix.c
> @@ -244,7 +244,9 @@ void qemu_set_nonblock(int fd)
>      f = fcntl(fd, F_GETFL);
>      assert(f != -1);
>      f = fcntl(fd, F_SETFL, f | O_NONBLOCK);
> +#ifndef __OpenBSD__
>      assert(f != -1);
> +#endif
>  }
>
>  int socket_set_fast_reuse(int fd)


--
Alex Bennée

On 1/28/19 10:47 AM, Alex Bennée wrote:
> 
> Philippe Mathieu-Daudé <philmd@redhat.com> writes:
> 
>> Previous to OpenBSD 6.3 [1], fcntl(F_SETFL) is not permitted on memory
>> devices. Do not assert fcntl failures on OpenBSD.
>> This fixes:
>>
>>   $ lm32-softmmu/qemu-system-lm32
>>   assertion "f != -1" failed: file "util/oslib-posix.c", line 247, function "qemu_set_nonblock"
>>   Abort trap (core dumped)
>>
>> [1] The fix seems https://github.com/openbsd/src/commit/c2a35b387f9d3c
>>   "fcntl(F_SETFL) invokes the FIONBIO and FIOASYNC ioctls internally, so
>>   the memory devices (/dev/null, /dev/zero, etc) need to permit them."
> 
> I assume set_nonblock is called on more than just these special devices?
> Is there anyway to check this on OpenBSD or is it just an anonymous fd
> at this point?

I'll let an OpenBSD expert to answer that.

I forgot to comment, the assert() was added one month ago (da93b82079d),
Michael said we can also revert this change.

>>
>> Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
>> ---
>>  util/oslib-posix.c | 2 ++
>>  1 file changed, 2 insertions(+)
>>
>> diff --git a/util/oslib-posix.c b/util/oslib-posix.c
>> index 4ce1ba9ca4..064c3ae2f7 100644
>> --- a/util/oslib-posix.c
>> +++ b/util/oslib-posix.c
>> @@ -244,7 +244,9 @@ void qemu_set_nonblock(int fd)
>>      f = fcntl(fd, F_GETFL);
>>      assert(f != -1);
>>      f = fcntl(fd, F_SETFL, f | O_NONBLOCK);
>> +#ifndef __OpenBSD__
>>      assert(f != -1);
>> +#endif
>>  }
>>
>>  int socket_set_fast_reuse(int fd)
> 
> 
> --
> Alex Bennée
> 

On 28/01/19 10:47, Alex Bennée wrote:
>>
>> [1] The fix seems https://github.com/openbsd/src/commit/c2a35b387f9d3c
>>   "fcntl(F_SETFL) invokes the FIONBIO and FIOASYNC ioctls internally, so
>>   the memory devices (/dev/null, /dev/zero, etc) need to permit them."
> I assume set_nonblock is called on more than just these special devices?
> Is there anyway to check this on OpenBSD or is it just an anonymous fd
> at this point?
> 

Perhaps on OpenBSD we should just assert that we don't get EBADF?

Paolo

On 1/28/19 12:03 PM, Paolo Bonzini wrote:
> On 28/01/19 10:47, Alex Bennée wrote:
>>>
>>> [1] The fix seems https://github.com/openbsd/src/commit/c2a35b387f9d3c
>>>   "fcntl(F_SETFL) invokes the FIONBIO and FIOASYNC ioctls internally, so
>>>   the memory devices (/dev/null, /dev/zero, etc) need to permit them."
>> I assume set_nonblock is called on more than just these special devices?
>> Is there anyway to check this on OpenBSD or is it just an anonymous fd
>> at this point?
>>
> 
> Perhaps on OpenBSD we should just assert that we don't get EBADF?

We get ENODEV for "not a memory device":

    19 ENODEV Operation not supported by device.
    An attempt was made to apply an inappropriate function to a device,
    for example, trying to read a write-only device such as a printer.

I'll respin with your suggestion.

Thanks!

Phil.