Date: Tue, 12 Nov 2013 13:48:35 -0800 From: Julian Elischer <julian@freebsd.org> To: Mateusz Guzik <mjguzik@gmail.com>, Sergey Kandaurov <pluknet@gmail.com> Cc: "freebsd-virtualization@freebsd.org" <freebsd-virtualization@freebsd.org> Subject: Re: Running `sysctl -a` causes crash (Hyper-V) Message-ID: <5282A233.9090401@freebsd.org> In-Reply-To: <20131112133135.GB29757@dft-labs.eu> References: <CAAoTqfvrdKH6BC=KrpWLgBKDjx402oTt_bp1TnV_Z3FC1if88A@mail.gmail.com> <CAE-mSOLKgk_h10mUPLuMXHgPfYJMEog_FFdyZvEoAOXMh=rbfg@mail.gmail.com> <CAAoTqfu9NFuVFYnLFo%2BLG7_kXd%2BAaxEPTm4nCKETzWvje7=fyA@mail.gmail.com> <CAE-mSOLeMHkh7gKLZPYcyu3GdJKqN4hAMBgEf5TwMUxB4nQ7zA@mail.gmail.com> <CAAoTqftAbsDNUiCv4QRnNfamf61dPhJ%2BWAzgfVVL4QMe9qHxcA@mail.gmail.com> <CAE-mSOLszJLfHBgmSTaUH5%2B5dTe23XLUFbp-nhfPpz2UVWMCJQ@mail.gmail.com> <CAAoTqfu716D-ifL-GXurxveZOKahLpndAivBWWt6ENHOJxXaaA@mail.gmail.com> <52814D25.1050603@freebsd.org> <20131112030154.GA29757@dft-labs.eu> <CAE-mSOJFGos7Y-QvuskPDK9=Jv8CLqGPKi8Gck=Uj2mmy87Y0Q@mail.gmail.com> <20131112133135.GB29757@dft-labs.eu>
next in thread | previous in thread | raw e-mail | index | archive | help
On 11/12/13, 5:31 AM, Mateusz Guzik wrote:
> On Tue, Nov 12, 2013 at 03:33:45PM +0400, Sergey Kandaurov wrote:
>> On 12 November 2013 07:01, Mateusz Guzik <mjguzik@gmail.com> wrote:
>>> On Mon, Nov 11, 2013 at 01:33:25PM -0800, Julian Elischer wrote:
>>>> On 11/11/13, 7:49 AM, Pavel Timofeev wrote:
>>>>> Great, it helped! Thanks a lot!
>>>> the suggested change suggests fragile code. It should be able to
>>>> handle unexpected input..
>>> Indeed, I already suggested Sergey placing a compile time assertion to
>>> make sure vm_guest range is covered.
>> I have the following change for this.
>> Btw, the last element in vm_guest_sysctl_names is NULL, is it really needed
>> for something? Unlike vm_bnames/vm_pnames, it seems not.
>>
> No idea, I would leave it be.
>
>> Add CTASSERT to handle enum VM_GUEST and vm_guest_sysctl_names cooperation.
>>
>> Index: sys/sys/systm.h
>> ===================================================================
>> --- sys/sys/systm.h (revision 258036)
>> +++ sys/sys/systm.h (working copy)
>> @@ -72,7 +72,8 @@ extern int vm_guest; /* Running as virtual machin
>> * ever implemented (e.g. vendor-specific paravirtualization features).
>> * Keep in sync with vm_guest_sysctl_names[].
>> */
>> -enum VM_GUEST { VM_GUEST_NO = 0, VM_GUEST_VM, VM_GUEST_XEN, VM_GUEST_HV };
>> +enum VM_GUEST { VM_GUEST_NO = 0, VM_GUEST_VM, VM_GUEST_XEN, VM_GUEST_HV,
>> + VM_LAST };
>>
>> #if defined(WITNESS) || defined(INVARIANTS)
>> void kassert_panic(const char *fmt, ...) __printflike(1, 2);
>> Index: sys/kern/subr_param.c
>> ===================================================================
>> --- sys/kern/subr_param.c (revision 258036)
>> +++ sys/kern/subr_param.c (working copy)
>> @@ -156,6 +156,7 @@ static const char *const vm_guest_sysctl_names[] =
>> "hv",
>> NULL
>> };
>> +CTASSERT(nitems(vm_guest_sysctl_names) - 1 == VM_LAST);
hmm. one learns something every day. nitems() is new to me..
yes assuming nitems is supported on all platforms then this adds a
safety belt
and is a good idea.
>>
>> #ifndef XEN
>> static const char *const vm_bnames[] = {
>>
> I can't test it right now, I presume it works :)
>
> With that in mind looks good to me, thanks!
>
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5282A233.9090401>