Date: Thu, 8 Dec 2016 19:55:41 +0100 From: Roman Divacky <rdivacky@vlakno.cz> To: Mark Millard <markmi@dsl-only.net> Cc: FreeBSD Toolchain <freebsd-toolchain@freebsd.org>, Justin Hibbits <chmeeedalf@gmail.com>, FreeBSD PowerPC ML <freebsd-ppc@freebsd.org> Subject: Re: clang 3.9.0 buildkernel on old powerpc64's vs. trying to build hwpmc_e500.o and the like. . . Message-ID: <20161208185541.GA33364@vlakno.cz> In-Reply-To: <CE88C1F4-B9BD-4D45-8DF0-F1079C3257A5@dsl-only.net> References: <300CB7A2-34BB-407F-B2E9-D263B119A989@dsl-only.net> <20161205161904.GA7889@vlakno.cz> <126E2EDE-9499-4103-A3DB-CC517105DAB2@dsl-only.net> <D3DE2D12-9885-4154-B680-6DA5A8B62A56@dsl-only.net> <D9C54972-8D21-4D55-A707-4FFC2BDCD9FE@dsl-only.net> <20161207190057.GA58950@vlakno.cz> <E1376C20-C1BD-418D-81C6-CDDE479342CA@dsl-only.net> <CE88C1F4-B9BD-4D45-8DF0-F1079C3257A5@dsl-only.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Can you try to investigate why it dies? I am not convinced -mminimal-toc
would result in a boot failure. I think the kernel would fail to link.
On Wed, Dec 07, 2016 at 09:52:47PM -0800, Mark Millard wrote:
> Top post of a FYI [head -r309656 powerpc64 context]:
>=20
> I commented out the one -mminimal-toc use in the modules and tried
> buildkernel again (cross build).
>=20
> It reached the end. But it dies immediately if I try to
> boot it after installing a copy.
>=20
> This was based on:
>=20
> # svnlite diff /usr/src/sys/modules/zfs/Makefile
> Index: /usr/src/sys/modules/zfs/Makefile
> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> --- /usr/src/sys/modules/zfs/Makefile (revision 309656)
> +++ /usr/src/sys/modules/zfs/Makefile (working copy)
> @@ -93,9 +93,9 @@
> CFLAGS+=3D-I${SUNW}/common
> CFLAGS+=3D-DBUILDING_ZFS
> =20
> -.if ${MACHINE_ARCH} =3D=3D "powerpc64"
> -CFLAGS+=3D-mminimal-toc
> -.endif
> +#.if ${MACHINE_ARCH} =3D=3D "powerpc64"
> +#CFLAGS+=3D-mminimal-toc
> +#.endif
> =20
> .ifdef ZFS_DEBUG
> CFLAGS+=3D-DDEBUG=3D1
>=20
> as well as your .td file patch.
>=20
> zfs is not in use in the configuration: it just
> uses ufs.
>=20
> I'll note that I had avoided 2.47 binutils variants
> based on reported issues in powerpc land (not that
> I know the details or the powerpc64 vs. powerpc vs.
> both status of the issues).
>=20
>=20
>=20
> =3D=3D=3D
> Mark Millard
> markmi at dsl-only.net
>=20
> On 2016-Dec-7, at 4:11 PM, Mark Millard <markmi@dsl-only.net> wrote:
>=20
> On 2016-Dec-7, at 11:00 AM, Roman Divacky <rdivacky at vlakno.cz> wrote:
>=20
> > Can the compiler you built with the patch process this file:
> >=20
> > typedef int register_t;
> > #define mtpmr(reg, val) =
\
> > __asm __volatile("mtpmr %0,%1" : : "K"(reg), "r"(val))
> > #define mfpmr(reg) =
\
> > ( { register_t val; \
> > __asm __volatile("mfpmr %0,%1" : "=3Dr"(val) : "K"(reg)); =
\
> > val; } )
> >=20
> > #define PMR_PMC0 16
> >=20
> > int foo()
> > {
> > return mfpmr(PMR_PMC0);
> > }
> >=20
> >=20
> > I can compile it just fine locally. Not sure why it wouldnt work in you=
r case.
>=20
> I separately had helped with testing for bugzilla 214902
> and so had updated to head -r309656 so the context was
> different.
>=20
> But I figured out the .td file related issue on powerpc64.
>=20
> My means of forcing all the compiles that target powerpc64
> to use -B to pick up the alternate toolchain (since the
> bootstrap binutils ld can fail) also forced the system
> compiler to be used instead of the bootstrapped clang.
> (The SRC_CONF_ENV file that I used had the text that
> forced this.)
>=20
> So my buildkernel was using an unpatched compiler when
> I tried kernel-toolchain then buildkernel.
>=20
> This was visible in the .meta file part of my report on the
> problem. It showed:
>=20
> > CMD /usr/bin/clang -B /usr/local/powerpc64-freebsd/bin/
>=20
>=20
> instead of having the path to the bootstrap compiler.
>=20
> It turns out that my amd64 cross build SRC_CONF_ENV file
> also had remnants of an experiment that also happened to
> force the system compiler to be used so it would have got
> the same behavior.
>=20
> Based on use of compilers that actually have your
> patch in them. . .
>=20
> Your patch worked fine to let the buildkernel reach
> the next problem: use of -mminimial-toc in a kernel
> module is made but is rejected for powerpc64.
>=20
> Sorry for the extra noise in reporting on your patch.
>=20
>=20
> Trying to find new things to report (future problems)
> by working around existing problems that are known but
> unfixed tends to have these sorts of interferences. Of
> course sometimes my workarounds might not be the best
> ones available.
>=20
> This stupid mistake is probably what is going on in
> at least one bugzilla report that I submitted: So
> I've likely got more to clean up.
>=20
> =3D=3D=3D
> Mark Millard
> markmi at dsl-only.net
>=20
> Older material. . .
>=20
> On Mon, Dec 05, 2016 at 05:42:28PM -0800, Mark Millard wrote:
> > On 2016-Dec-5, at 5:16 PM, Mark Millard <markmi at dsl-only.net> wrote:
> >=20
> >> Well it looks like:
> >>=20
> >> WITHOUT_CROSS_COMPILER=3D
> >> WITH_SYSTEM_COMPILER=3D
> >>=20
> >> ignores the .td file change but
> >>=20
> >> WITH_CROSS_COMPILER=3D
> >> WITHOUT_SYSTEM_COMPILER=3D
> >>=20
> >> may use it.
> >>=20
> >> I had accidentally used a SRC_CONF_ENV file that
> >> was of the first form.
> >>=20
> >> So I've got a build going based on the 2nd form. . .
> >=20
> > No such luck: same type of failure at the same point.
> >=20
> > =3D=3D=3D
> > Mark Millard
> > markmi at dsl-only.net
> >=20
> > On 2016-Dec-5, at 4:05 PM, Mark Millard <markmi at dsl-only.net> wrote:
> >=20
> > On 2016-Dec-5, at 8:19 AM, Roman Divacky <rdivacky at vlakno.cz> wrote:
> >=20
> >> Can you try this patch?
> >>=20
> >> Index: llvm/lib/Target/PowerPC/PPCInstrInfo.td
> >> =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> >> --- llvm/lib/Target/PowerPC/PPCInstrInfo.td (revision 288675)
> >> +++ llvm/lib/Target/PowerPC/PPCInstrInfo.td (working copy)
> >> @@ -2373,6 +2373,12 @@
> >> def MFTB : XFXForm_1<31, 371, (outs gprc:$RT), (ins i32imm:$SPR),
> >> "mftb $RT, $SPR", IIC_SprMFTB>;
> >>=20
> >> +def MFPMR : XFXForm_1<31, 334, (outs gprc:$RT), (ins i32imm:$PMRN),
> >> + "mfpmr $RT, $PMRN", IIC_IntGeneral>;
> >> +
> >> +def MTPMR : XFXForm_1<31, 462, (outs), (ins i32imm:$PMRN, gprc:$RS),
> >> + "mtpmr $PMRN, $RS", IIC_IntGeneral>;
> >> +
> >> // A pseudo-instruction used to implement the read of the 64-bit cycle=
counter
> >> // on a 32-bit target.
> >> let hasSideEffects =3D 1, usesCustomInserter =3D 1 in
> >=20
> > Direct use of the patch (put into a file) was rejected:
> >=20
> > # patch -p0 < llvmPPCInstrInfo_td.patch=20
> > Hmm... Looks like a unified diff to me...
> > The text leading up to this was:
> > --------------------------
> > |Index: llvm/lib/Target/PowerPC/PPCInstrInfo.td
> > |=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> > |--- llvm/lib/Target/PowerPC/PPCInstrInfo.td (revision 288675)
> > |+++ llvm/lib/Target/PowerPC/PPCInstrInfo.td (working copy)
> > --------------------------
> > Patching file llvm/lib/Target/PowerPC/PPCInstrInfo.td using Plan A...
> > patch: **** malformed patch at line 6: def MFTB : XFXForm_1<31, 371, (o=
uts gprc:$RT), (ins i32imm:$SPR),
> >=20
> > So I hand put in the extra lines.
> >=20
> > I'll note that in llvm/lib/Target/PowerPC/PPCInstrInfo.td -r309124
> > the MFTB line is at line number 2300 while your patch listed:
> >=20
> > @@ -2373,6 +2373,12 @@
> >=20
> > My edit shows as:
> >=20
> > # svnlite diff contrib/llvm/lib/Target/PowerPC/PPCInstrInfo.td
> > Index: contrib/llvm/lib/Target/PowerPC/PPCInstrInfo.td
> > =3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D
> > --- contrib/llvm/lib/Target/PowerPC/PPCInstrInfo.td (revision 309179)
> > +++ contrib/llvm/lib/Target/PowerPC/PPCInstrInfo.td (working copy)
> > @@ -2300,6 +2300,12 @@
> > def MFTB : XFXForm_1<31, 371, (outs gprc:$RT), (ins i32imm:$SPR),
> > "mftb $RT, $SPR", IIC_SprMFTB>;
> >=20
> > +def MFPMR : XFXForm_1<31, 334, (outs gprc:$RT), (ins i32imm:$PMRN),
> > + "mfpmr $RT, $PMRN", IIC_IntGeneral>;
> > +
> > +def MTPMR : XFXForm_1<31, 462, (outs), (ins i32imm:$PMRN, gprc:$RS),
> > + "mtpmr $PMRN, $RS", IIC_IntGeneral>;
> > +
> > // A pseudo-instruction used to implement the read of the 64-bit cycle =
counter
> > // on a 32-bit target.
> > let hasSideEffects =3D 1, usesCustomInserter =3D 1 in
> >=20
> >=20
> > Unfortunately the buildkernel still gets the same errors:
> > (This was tried after a kernel-toolchain .)
> >=20
> > # Meta data file /usr/obj/powerpc64vtsc_clang_altbinutils_kernel/powerp=
c.powerpc64/usr/src/sys/GENERIC64vtsc-NODBG/modules/usr/src/sys/modules/hwp=
mc/hwpmc_e500.o.meta
> > CMD /usr/bin/clang -B /usr/local/powerpc64-freebsd/bin/ -target powerpc=
64-unknown-freebsd12.0 --sysroot=3D/usr/obj/powerpc64vtsc_clang_altbinutils=
_kernel/powerpc.powerpc64/usr/src/tmp -B/usr/local/powerpc64-freebsd/bin/ -=
O2 -pipe -fno-strict-aliasing -D_KERNEL -DKLD_MODULE -nostdinc -DHAVE_KE=
RNEL_OPTION_HEADERS -include /usr/obj/powerpc64vtsc_clang_altbinutils_kerne=
l/powerpc.powerpc64/usr/src/sys/GENERIC64vtsc-NODBG/opt_global.h -I. -I/usr=
/src/sys -fno-common -g -fno-omit-frame-pointer -I/usr/obj/powerpc64vtsc_cl=
ang_altbinutils_kernel/powerpc.powerpc64/usr/src/sys/GENERIC64vtsc-NODBG -=
mno-altivec -ffreestanding -fwrapv -fstack-protector -gdwarf-2 -Wall -Wredu=
ndant-decls -Wnested-externs -Wstrict-prototypes -Wmissing-prototypes -Wpoi=
nter-arith -Winline -Wcast-qual -Wundef -Wno-pointer-sign -D__printf__=3D__=
freebsd_kprintf__ -Wmissing-include-dirs -fdiagnostics-show-option -Wno-unk=
nown-pragmas -Wno-error-tautological-compare -Wno-error-empty-body -Wno-err=
or-parentheses-equality -Wno-e
> > rror-unused-function -Wno-error-pointer-sign -Wno-error-shift-negative-=
value -msoft-float -std=3Diso9899:1999 -c /usr/src/sys/modules/hwpmc/../.=
=2E/dev/hwpmc/hwpmc_e500.c -o hwpmc_e500.o
> > CMD ctfconvert -L VERSION -g hwpmc_e500.o
> > CWD /usr/obj/powerpc64vtsc_clang_altbinutils_kernel/powerpc.powerpc64/u=
sr/src/sys/GENERIC64vtsc-NODBG/modules/usr/src/sys/modules/hwpmc
> > TARGET hwpmc_e500.o
> > -- command output --
> > /usr/src/sys/modules/hwpmc/../../dev/hwpmc/hwpmc_e500.c:475:19: error: =
unrecognized instruction mnemonic
> > uint32_t pmgc0 =3D mfpmr(PMR_PMGC0);
> > ^
> > ./machine/pmc_mdep.h:24:21: note: expanded from macro 'mfpmr'
> > __asm __volatile("mfpmr %0,%1" : "=3Dr"(val) : "K"(reg)); \
> > ^
> > <inline asm>:1:2: note: instantiated into assembly here
> > mfpmr 3,400
> > ^
> > /usr/src/sys/modules/hwpmc/../../dev/hwpmc/hwpmc_e500.c:478:2: error: u=
nrecognized instruction mnemonic
> > mtpmr(PMR_PMGC0, pmgc0);
> > ^
> > ./machine/pmc_mdep.h:21:19: note: expanded from macro 'mtpmr'
> > __asm __volatile("mtpmr %0,%1" : : "K"(reg), "r"(val))
> > ^
> > <inline asm>:1:2: note: instantiated into assembly here
> > mtpmr 400,3
> > ^
> > /usr/src/sys/modules/hwpmc/../../dev/hwpmc/hwpmc_e500.c:446:2: error: u=
nrecognized instruction mnemonic
> > mtpmr(PMR_PMGC0, PMGC_FAC | PMGC_PMIE | PMGC_FCECE);
> > ^
> > ./machine/pmc_mdep.h:21:19: note: expanded from macro 'mtpmr'
> > __asm __volatile("mtpmr %0,%1" : : "K"(reg), "r"(val))
> > ^
> > <inline asm>:1:2: note: instantiated into assembly here
> > mtpmr 400,3
> > ^
> > /usr/src/sys/modules/hwpmc/../../dev/hwpmc/hwpmc_e500.c:408:14: error: =
unrecognized instruction mnemonic
> > pmc_pmlc =3D mfpmr(PMR_PMLCa0);
> > ^
> > ./machine/pmc_mdep.h:24:21: note: expanded from macro 'mfpmr'
> > __asm __volatile("mfpmr %0,%1" : "=3Dr"(val) : "K"(reg)); \
> > ^
> > <inline asm>:1:2: note: instantiated into assembly here
> > mfpmr 10,144
> > ^
> > . . .
> >=20
> >=20
> > =3D=3D=3D
> > Mark Millard
> > markmi at dsl-only.net
> >=20
> > Older content:
> >=20
> > On Sat, Dec 03, 2016 at 08:35:50PM -0800, Mark Millard wrote:
> >> [Note: At present I can buildworld using WITH_LIB32=3D on
> >> powerpc64 for TARGET_ARCH=3Dpowerpc64 via clang 3.9.0 from a
> >> minor variant of head -r309179 . That does not work for
> >> amd64 cross compiling to powerpc64 due to assembler
> >> notation rejections.]
> >>=20
> >> When I attempt to buildkernel (with WERROR=3D ) via FreeBSD's
> >> clang 3.9.0 I get the following sorts of error
> >> reports, *even building on powerpc64* :
> >>=20
> >> --- hwpmc_e500.o ---
> >> /usr/src/sys/modules/hwpmc/../../dev/hwpmc/hwpmc_e500.c:475:19: error:=
unrecognized instruction mnemonic
> >> uint32_t pmgc0 =3D mfpmr(PMR_PMGC0);
> >> ^
> >> ./machine/pmc_mdep.h:24:21: note: expanded from macro 'mfpmr'
> >> __asm __volatile("mfpmr %0,%1" : "=3Dr"(val) : "K"(reg)); \
> >> ^
> >> <inline asm>:1:2: note: instantiated into assembly here
> >> mfpmr 3,400
> >> ^
> >> . . .
> >>=20
> >> When I look up these instructions I find that they are not
> >> classic powerpc instructions:
> >> ( http://www.nxp.com/assets/documents/data/en/white-papers/POWRPCARCPR=
MRM.pdf )
> >>=20
> >>> Whereas the classic architecture defines special-purpose registers (S=
PRs) and
> >>> two instructions to access them (Move to Special-Purpose Register (mt=
spr) and
> >>> Move from Special-Purpose Register (mfspr)), Book E takes that model =
and defines
> >>> optional device control registers (DCRs) and mtdcr and mfdcr instruct=
ions, and
> >>> the EIS-defined performance monitor APU defines performance monitor r=
egisters
> >>> (PMRs) and mtpmr and mfpmr instructions, all based on models provided=
by the
> >>> UISA.
> >>=20
> >> . . .
> >>=20
> >> Does this imply that clang 3.9.0 needs to support more instructions wh=
en
> >> it is targeting FreeBSD for GENERIC64 based builds? (I include GENERIC=
64
> >> and then override some items for my build activity.)
> >>=20
> >> If yes, then someone probably needs to make a list of what instructions
> >> need to be present and have someone submit the list into llvm's bugzil=
la
> >> and have the submittal listed in:
> >>=20
> >> [META] Using Clang as the FreeBSD/ppc system compiler
> >>=20
> >> (25780).
> >>=20
> >>=20
> >> If GENERIC64 does not need the likes of hwpmc_e500.o built then some
> >> work on the build environment to avoid GENERIC64 including things with
> >> non-classic instruction use would appear to be needed. (No llvm
> >> involvement for this case.) I doubt this is the case as it would
> >> seem that the problem would reoccur when alternate KERNCONF's were
> >> in use instead that do require something like of hwpmc_e500.o to be
> >> built.
> >>=20
> >>=20
> >> Note: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D214903 is ab=
out
> >> this issue from the FreeBSD side of things. I just noticed the status
> >> of the specific instructions involved and also that the cross-build
> >> and on-powerpc64 build get the same result (unlike the recent
> >> WITH_LIB32=3D discovery).
> >>=20
> >> =3D=3D=3D
> >> Mark Millard
> >> markmi at dsl-only.net
> >=20
> >=20
> > _______________________________________________
> > freebsd-ppc@freebsd.org mailing list
> > https://lists.freebsd.org/mailman/listinfo/freebsd-ppc
> > To unsubscribe, send any mail to "freebsd-ppc-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20161208185541.GA33364>