qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH 1/3] linux-user: implement FS_IOC_GETFLAGS ioctl


From: Alexander Graf
Subject: Re: [Qemu-devel] [PATCH 1/3] linux-user: implement FS_IOC_GETFLAGS ioctl
Date: Thu, 11 Oct 2012 15:52:17 +0200

On 11.10.2012, at 15:36, Riku Voipio wrote:

> Hi Alexander,
> 
> On 21 August 2012 16:47, Peter Maydell <address@hidden> wrote:
>>> +#define TARGET_FS_IOC_GETFLAGS TARGET_IORU('f', 1)
>> 
>> This and the SETFLAGS one in the next patch fail the consistency
>> check that an x86_64-on-x86_64 linux-user binary performs:
>> 
>> cam-vm-266:precise:qemu$ ./x86_64-linux-user/qemu-x86_64 /bin/echo hello
>> ERROR: ioctl(FS_IOC_GETFLAGS): target=0x80046601 host=0x80086601
>> ERROR: ioctl(FS_IOC_SETFLAGS): target=0x40046602 host=0x40086602
>> hello
>> 
>> This is indicating that your ioctl definition is wrong:
>>> +  IOCTL(FS_IOC_GETFLAGS, IOC_R, MK_PTR(TYPE_INT))
>> 
>> ...it should be TYPE_LONG.
>> 
>> Incidentally you could also set the target ioctl at compile
>> time rather than making syscall_init patch the size field
>> at runtime. I don't know whether one or the other is better
>> style... There's an argument for the 'automatic' one as
>> it enforces consistency and catches errors like the one above.
>> Anyway, the compile-time option woud be:
>> 
>> #define TARGET_FS_IOC_GETFLAGS TARGET_IOR('f', 1, abi_long)
>> 
>> (You could also squash patches 1 and 2 together IMHO.)
> 
> Has there been an updated patch with Peter's suggestions sent the list since?

Didn't get around to it yet, no.


Alex

> 
> Riku




reply via email to

[Prev in Thread] Current Thread [Next in Thread]