| Message ID | 1559272702-13459-1-git-send-email-chenbaodong@mxnavi.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | xen: put cpupool's member 'n_dom' after 'cpupool_id' | expand |
Hi, On 5/31/19 4:18 AM, Baodong Chen wrote: > Thus, sizeof(struct cpupool) will save 8 bytes for 64-bit system. I am happy with the change, although AFAIK cpupool is not instantiated that often. Are you planning to have more instantiation of it? Cheers, > > Signed-off-by: Baodong Chen <chenbaodong@mxnavi.com> > --- > xen/include/xen/sched-if.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/xen/include/xen/sched-if.h b/xen/include/xen/sched-if.h > index 92bc7a0..f0cf210 100644 > --- a/xen/include/xen/sched-if.h > +++ b/xen/include/xen/sched-if.h > @@ -213,9 +213,9 @@ static inline void sched_free_domdata(const struct scheduler *s, > struct cpupool > { > int cpupool_id; > + unsigned int n_dom; > cpumask_var_t cpu_valid; /* all cpus assigned to pool */ > struct cpupool *next; > - unsigned int n_dom; > struct scheduler *sched; > atomic_t refcnt; > }; >
On 5/31/19 18:52, Julien Grall wrote: > Hi, > > On 5/31/19 4:18 AM, Baodong Chen wrote: >> Thus, sizeof(struct cpupool) will save 8 bytes for 64-bit system. > > I am happy with the change, although AFAIK cpupool is not instantiated > that often. Are you planning to have more instantiation of it? > > Cheers, > No, I'm not planning to create lots of cpupool instance. I'm studying xen for a few weeks and my plan is: run xen for embeded automotive use case with dom0less, null scheduler, small code base for safety certified maybe a plus(not sure whether can do this). >> >> Signed-off-by: Baodong Chen <chenbaodong@mxnavi.com> >> --- >> xen/include/xen/sched-if.h | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/xen/include/xen/sched-if.h b/xen/include/xen/sched-if.h >> index 92bc7a0..f0cf210 100644 >> --- a/xen/include/xen/sched-if.h >> +++ b/xen/include/xen/sched-if.h >> @@ -213,9 +213,9 @@ static inline void sched_free_domdata(const >> struct scheduler *s, >> struct cpupool >> { >> int cpupool_id; >> + unsigned int n_dom; >> cpumask_var_t cpu_valid; /* all cpus assigned to pool */ >> struct cpupool *next; >> - unsigned int n_dom; >> struct scheduler *sched; >> atomic_t refcnt; >> }; >> >
Hi, On 03/06/2019 02:48, chenbaodong wrote: > > On 5/31/19 18:52, Julien Grall wrote: >> Hi, >> >> On 5/31/19 4:18 AM, Baodong Chen wrote: >>> Thus, sizeof(struct cpupool) will save 8 bytes for 64-bit system. >> >> I am happy with the change, although AFAIK cpupool is not instantiated that >> often. Are you planning to have more instantiation of it? >> >> Cheers, >> > No, I'm not planning to create lots of cpupool instance. > > I'm studying xen for a few weeks and my plan is: > > run xen for embeded automotive use case with dom0less, null scheduler, small > code base for safety certified maybe a plus(not sure whether can do this). FWIW, there are discussion to get Xen safety certified. They are captured on [1]. Cheers, [1] https://wiki.xenproject.org/wiki/Category:Safety_Certification/FuSa_SIG > > >>> >>> Signed-off-by: Baodong Chen <chenbaodong@mxnavi.com> >>> --- >>> xen/include/xen/sched-if.h | 2 +- >>> 1 file changed, 1 insertion(+), 1 deletion(-) >>> >>> diff --git a/xen/include/xen/sched-if.h b/xen/include/xen/sched-if.h >>> index 92bc7a0..f0cf210 100644 >>> --- a/xen/include/xen/sched-if.h >>> +++ b/xen/include/xen/sched-if.h >>> @@ -213,9 +213,9 @@ static inline void sched_free_domdata(const struct >>> scheduler *s, >>> struct cpupool >>> { >>> int cpupool_id; >>> + unsigned int n_dom; >>> cpumask_var_t cpu_valid; /* all cpus assigned to pool */ >>> struct cpupool *next; >>> - unsigned int n_dom; >>> struct scheduler *sched; >>> atomic_t refcnt; >>> }; >>> >>
On 6/3/19 17:30, Julien Grall wrote: > Hi, > > On 03/06/2019 02:48, chenbaodong wrote: >> >> On 5/31/19 18:52, Julien Grall wrote: >>> Hi, >>> >>> On 5/31/19 4:18 AM, Baodong Chen wrote: >>>> Thus, sizeof(struct cpupool) will save 8 bytes for 64-bit system. >>> >>> I am happy with the change, although AFAIK cpupool is not >>> instantiated that often. Are you planning to have more instantiation >>> of it? >>> >>> Cheers, >>> >> No, I'm not planning to create lots of cpupool instance. >> >> I'm studying xen for a few weeks and my plan is: >> >> run xen for embeded automotive use case with dom0less, null >> scheduler, small code base for safety certified maybe a plus(not sure >> whether can do this). > > FWIW, there are discussion to get Xen safety certified. They are > captured on [1]. > > Cheers, > > [1] > https://wiki.xenproject.org/wiki/Category:Safety_Certification/FuSa_SIG Hello Julien, Thanks for your info. I will follow the wiki link and see what i can do about safety certification. > >> >> >>>> >>>> Signed-off-by: Baodong Chen <chenbaodong@mxnavi.com> >>>> --- >>>> xen/include/xen/sched-if.h | 2 +- >>>> 1 file changed, 1 insertion(+), 1 deletion(-) >>>> >>>> diff --git a/xen/include/xen/sched-if.h b/xen/include/xen/sched-if.h >>>> index 92bc7a0..f0cf210 100644 >>>> --- a/xen/include/xen/sched-if.h >>>> +++ b/xen/include/xen/sched-if.h >>>> @@ -213,9 +213,9 @@ static inline void sched_free_domdata(const >>>> struct scheduler *s, >>>> struct cpupool >>>> { >>>> int cpupool_id; >>>> + unsigned int n_dom; >>>> cpumask_var_t cpu_valid; /* all cpus assigned to pool */ >>>> struct cpupool *next; >>>> - unsigned int n_dom; >>>> struct scheduler *sched; >>>> atomic_t refcnt; >>>> }; >>>> >>> >
>>> On 31.05.19 at 05:18, <chenbaodong@mxnavi.com> wrote: > Thus, sizeof(struct cpupool) will save 8 bytes for 64-bit system. > > Signed-off-by: Baodong Chen <chenbaodong@mxnavi.com> > --- > xen/include/xen/sched-if.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) I've just noticed that this file isn't part of the SCHEDULING maintainer group. Therefore Acked-by: Jan Beulich <jbeulich@suse.com> I'm wondering though whether maintainership of the file shouldn't be adjusted, preferably by moving this (supposedly private) header into xen/common/. I realize there may be some further cleanup needed for this to actually be possible. If there was general consensus about such a step, I could see about actually carrying it out. Jan
On 17.06.19 17:19, Jan Beulich wrote: >>>> On 31.05.19 at 05:18, <chenbaodong@mxnavi.com> wrote: >> Thus, sizeof(struct cpupool) will save 8 bytes for 64-bit system. >> >> Signed-off-by: Baodong Chen <chenbaodong@mxnavi.com> >> --- >> xen/include/xen/sched-if.h | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) > > I've just noticed that this file isn't part of the SCHEDULING > maintainer group. Therefore > > Acked-by: Jan Beulich <jbeulich@suse.com> > > I'm wondering though whether maintainership of the file shouldn't be > adjusted, preferably by moving this (supposedly private) header into > xen/common/. I realize there may be some further cleanup needed > for this to actually be possible. If there was general consensus about > such a step, I could see about actually carrying it out. I am planning a big scheduling cleanup series after my core scheduling series has been processed. sched-if.h is part of that work. Juergen
diff --git a/xen/include/xen/sched-if.h b/xen/include/xen/sched-if.h index 92bc7a0..f0cf210 100644 --- a/xen/include/xen/sched-if.h +++ b/xen/include/xen/sched-if.h @@ -213,9 +213,9 @@ static inline void sched_free_domdata(const struct scheduler *s, struct cpupool { int cpupool_id; + unsigned int n_dom; cpumask_var_t cpu_valid; /* all cpus assigned to pool */ struct cpupool *next; - unsigned int n_dom; struct scheduler *sched; atomic_t refcnt; };
Thus, sizeof(struct cpupool) will save 8 bytes for 64-bit system. Signed-off-by: Baodong Chen <chenbaodong@mxnavi.com> --- xen/include/xen/sched-if.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)