[8u] RFR: JDK-8164041: support old pre-c++11 toolchains and ucLibc

[Andrey Petushkov]

Hmm, interesting. I will investigate. There are 3 "*eq" instructions to
follow hence there should be 3 additional "t"s after "it"

Anyway, please don't bother now, have a good time there! :)

Andrey

On Wed, Aug 24, 2016 at 7:09 PM Alex Kashchenko <akashche at redhat.com> wrote:

> Hi Andrey,
>
> On 08/24/2016 06:25 PM, Andrey Petushkov wrote:
> > Hi Alex,
> >
> > You, you're right and thank you for the fix! One small correction, I
> > believe there should be "itttt" rather than "ittt" :)
>
> I believe it is "ittt" (sorry, cannot check right now, currently on
> vacation), as I remember jvm hangs on startup with "itttt". But of
> course this is a very minor detail that will be immediately clear from
> thumb builds.
>
> >
> > I'll include this fix into second revision of the patch update in
> > accordance with the results of discussion with Ed
> >
> > Regards,
> > Andrey
> >
> > PS. Pardon for delay, was busy preparing the C1 monster, it did not come
> > easy for some reason
> >
> > On Fri, Aug 19, 2016 at 9:51 AM Alex Kashchenko <akashche at redhat.com>
> wrote:
> >
> >> Hi,
> >>
> >> On 08/15/2016 07:06 PM, Andrey Petushkov wrote:
> >>> Hi All,
> >>>
> >>> We met a necessity to support an older kind of gcc toolchain which does
> >> not fully conform to c++ 11 and in particular do not have __atomic API
> as
> >> well as have some other peculiarities. In addition the support of ucLibc
> >> was desired.
> >>>
> >>> The resulting changes are here:
> >> http://cr.openjdk.java.net/~apetushkov/8164041/webrev/ <
> >> http://cr.openjdk.java.net/~apetushkov/8164041/webrev/>
> >>> Checked with running on OpenWRT-based distribution
> >>>
> >>> When implementing the replacement code we also chosen to always use the
> >> assembly implementation and not builtins because:
> >>> - there is no harm, the assembly is no worse than that gcc can generate
> >>> - the code is more clear
> >>> - there is no reliable way to determine if gcc has flaws or not, so the
> >> detection directives look really ugly
> >>>
> >>> Thanks,
> >>> Andrey
> >>>
> >>
> >> A small fix for thumb mode (that is default on ubuntu xenial): in
> >> Atomic::cmpxchg add "ittt eq" before the "cmpeq". Otherwise it will fail
> >> on thumb with:
> >>
> >> {standard input}: Assembler messages:
> >> {standard input}:926: Error: thumb conditional instruction should be in
> >> IT block -- `cmpeq r6,r2'
> >> {standard input}:927: Error: thumb conditional instruction should be in
> >> IT block -- `strexdeq lr,r3,r4,[r0]'
> >> {standard input}:928: Error: thumb conditional instruction should be in
> >> IT block -- `cmpeq lr,#1'
> >> /home/u1604/jdk8u/hotspot/make/linux/makefiles/rules.make:157: recipe
> >> for target 'atomic.o' failed
> >>
> >>
> >> --
> >> -Alex
> >>
> >
>
>
> --
> -Alex
>