Date: Tue, 30 Jul 2019 20:40:28 -0700 From: Bryan Drewery <bdrewery@FreeBSD.org> To: Konstantin Belousov <kostikbel@gmail.com> Cc: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Subject: Re: svn commit: r350005 - head/sys/kern Message-ID: <5d3ffe05-58f3-ea26-13ed-51cdcd1deeaa@FreeBSD.org> In-Reply-To: <20190730232714.GM2731@kib.kiev.ua> References: <201907151918.x6FJIPFo077975@repo.freebsd.org> <d281e1dc-1931-20a5-9f41-7c1f66d4f667@FreeBSD.org> <20190730231256.GL2731@kib.kiev.ua> <20190730232714.GM2731@kib.kiev.ua>
next in thread | previous in thread | raw e-mail | index | archive | help
This is an OpenPGP/MIME signed message (RFC 4880 and 3156) --hwljxk2k8uM2tHhElvSoAgtwMvoOhwLgC Content-Type: multipart/mixed; boundary="p9Fqr5lLFAnCGEL5t8G4mRlXehaxggNpR"; protected-headers="v1" From: Bryan Drewery <bdrewery@FreeBSD.org> To: Konstantin Belousov <kostikbel@gmail.com> Cc: src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org Message-ID: <5d3ffe05-58f3-ea26-13ed-51cdcd1deeaa@FreeBSD.org> Subject: Re: svn commit: r350005 - head/sys/kern References: <201907151918.x6FJIPFo077975@repo.freebsd.org> <d281e1dc-1931-20a5-9f41-7c1f66d4f667@FreeBSD.org> <20190730231256.GL2731@kib.kiev.ua> <20190730232714.GM2731@kib.kiev.ua> In-Reply-To: <20190730232714.GM2731@kib.kiev.ua> --p9Fqr5lLFAnCGEL5t8G4mRlXehaxggNpR Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: quoted-printable On 7/30/19 4:27 PM, Konstantin Belousov wrote: > On Wed, Jul 31, 2019 at 02:13:02AM +0300, Konstantin Belousov wrote: >> On Tue, Jul 30, 2019 at 03:04:57PM -0700, Bryan Drewery wrote: >>> On 7/15/2019 12:18 PM, Konstantin Belousov wrote: >>>> Author: kib >>>> Date: Mon Jul 15 19:18:25 2019 >>>> New Revision: 350005 >>>> URL: https://svnweb.freebsd.org/changeset/base/350005 >>>> >>>> Log: >>>> In do_sem2_wait(), balance umtx_key_get() with umtx_key_release() = on retry. >>>> =20 >>> >>> Is this also needed in do_sem_wait()? A similar pattern seems to be t= here. >> No, I do not think do_sem_wait() has similar issue, because the again = label >> does not re-get the key. >> >>> >>> I ask because of what I referenced on IRC. I have some processes stuc= k >>> in here from a 10.4 jail. >>> >>>> ~/git/poudriere # procstat -kk 1498 >>>> PID TID COMM TDNAME KSTACK >>>> 1498 100710 python2.7 - mi_switch+0x174= sleepq_switch+0x110 sleepq_catch_signals+0x417 sleepq_wait_sig+0xf _slee= p+0x2d0 umtxq_sleep+0x153 do_sem_wait+0x42c __umtx_op_sem_wait+0x6e amd64= _syscall+0x2bb fast_syscall_common+0x101 >>>> 1498 101575 python2.7 - mi_switch+0x174= sleepq_switch+0x110 sleepq_catch_signals+0x417 sleepq_wait_sig+0xf _slee= p+0x2d0 umtxq_sleep+0x153 do_sem_wait+0x42c __umtx_op_sem_wait+0x6e amd64= _syscall+0x2bb fast_syscall_common+0x101 >>>> 1498 101657 python2.7 - <running> >>> ... >>>> ~/git/poudriere # procstat -kk 1498 >>>> PID TID COMM TDNAME KSTACK >>>> 1498 100710 python2.7 - mi_switch+0x174= sleepq_switch+0x110 sleepq_catch_signals+0x417 sleepq_wait_sig+0xf _slee= p+0x2d0 umtxq_sleep+0x153 do_sem_wait+0x42c __umtx_op_sem_wait+0x6e amd64= _syscall+0x2bb fast_syscall_common+0x101 >>>> 1498 101575 python2.7 - mi_switch+0x174= sleepq_switch+0x110 sleepq_catch_signals+0x417 sleepq_wait_sig+0xf _slee= p+0x2d0 umtxq_sleep+0x153 do_sem_wait+0x42c __umtx_op_sem_wait+0x6e amd64= _syscall+0x2bb fast_syscall_common+0x101 >>>> 1498 101657 python2.7 - do_sem_wait+0x1= b6 __umtx_op_sem_wait+0x6e amd64_syscall+0x2bb fast_syscall_common+0x101 >>> ... >>>> ~/git/poudriere # procstat -kk 94392 >>>> PID TID COMM TDNAME KSTACK >>>> 94392 101815 python2.7 - mi_switch+0x174= sleepq_switch+0x110 sleepq_catch_signals+0x417 sleepq_wait_sig+0xf _slee= p+0x2d0 umtxq_sleep+0x153 do_sem_wait+0x42c __umtx_op_sem_wait+0x6e amd64= _syscall+0x2bb fast_syscall_common+0x101 >>>> 94392 101816 python2.7 - __mtx_lock_slee= p+0x118 __mtx_lock_flags+0x102 _sleep+0x334 umtxq_busy+0xb7 do_sem_wait+0= x161 __umtx_op_sem_wait+0x6e amd64_syscall+0x2bb fast_syscall_common+0x10= 1 >>>> 94392 102076 python2.7 - __mtx_lock_flag= s+0x94 do_sem_wait+0x228 __umtx_op_sem_wait+0x6e amd64_syscall+0x2bb fast= _syscall_common+0x101 >> >> Try this. We should only retry casueword if it failed spuriously. >> >> diff --git a/sys/kern/kern_umtx.c b/sys/kern/kern_umtx.c >> index bb998457975..6c914ab6f3e 100644 >> --- a/sys/kern/kern_umtx.c >> +++ b/sys/kern/kern_umtx.c >> @@ -3229,7 +3229,8 @@ do_sem_wait(struct thread *td, struct _usem *sem= , struct _umtx_time *timeout) >> rv =3D casueword32(&sem->_has_waiters, 0, &count1, 1); >> if (rv =3D=3D 0) >> rv1 =3D fueword32(&sem->_count, &count); >> - if (rv =3D=3D -1 || (rv =3D=3D 0 && (rv1 =3D=3D -1 || count !=3D 0))= || rv =3D=3D 1) { >> + if (rv =3D=3D -1 || (rv =3D=3D 0 && (rv1 =3D=3D -1 || count !=3D 0))= || >> + (rv =3D=3D 1 && count1 =3D=3D 0)) { >> umtxq_lock(&uq->uq_key); >> umtxq_unbusy(&uq->uq_key); >> umtxq_remove(uq); >=20 > I think there is another problem, since even despite our intent of loop= ing > just because of casueword returned 1, the umtxq_check_susp() should hav= e > terminated the loop. I believe the following update would fix that. >=20 > If you have time, can you please apply only the umtxq_check_susp() chun= k > and see if it helps as well ? [Both chunks are needed for correctness, = but > the umtxq_check_susp() is almost impossible to test in combination] >=20 > diff --git a/sys/kern/kern_umtx.c b/sys/kern/kern_umtx.c > index bb998457975..08bdd1a1a9a 100644 > --- a/sys/kern/kern_umtx.c > +++ b/sys/kern/kern_umtx.c > @@ -723,13 +723,11 @@ umtxq_check_susp(struct thread *td, bool sleep) > error =3D 0; > p =3D td->td_proc; > PROC_LOCK(p); > - if (P_SHOULDSTOP(p) || > - ((p->p_flag & P_TRACED) && (td->td_dbgflags & TDB_SUSPEND))) { > - if (p->p_flag & P_SINGLE_EXIT) > - error =3D EINTR; > - else > - error =3D sleep ? thread_suspend_check(0) : ERESTART; > - } > + if (p->p_flag & P_SINGLE_EXIT) > + error =3D EINTR; > + else if (P_SHOULDSTOP(p) || > + ((p->p_flag & P_TRACED) && (td->td_dbgflags & TDB_SUSPEND))) > + error =3D sleep ? thread_suspend_check(0) : ERESTART; > PROC_UNLOCK(p); > return (error); > } > @@ -3229,7 +3227,8 @@ do_sem_wait(struct thread *td, struct _usem *sem,= struct _umtx_time *timeout) > rv =3D casueword32(&sem->_has_waiters, 0, &count1, 1); > if (rv =3D=3D 0) > rv1 =3D fueword32(&sem->_count, &count); > - if (rv =3D=3D -1 || (rv =3D=3D 0 && (rv1 =3D=3D -1 || count !=3D 0)) = || rv =3D=3D 1) { > + if (rv =3D=3D -1 || (rv =3D=3D 0 && (rv1 =3D=3D -1 || count !=3D 0)) = || > + (rv =3D=3D 1 && count1 =3D=3D 0)) { > umtxq_lock(&uq->uq_key); > umtxq_unbusy(&uq->uq_key); > umtxq_remove(uq); >=20 This 2nd change alone (&& count1 =3D=3D 0) was sufficient to fix the endl= ess loop problem. I am not sure how to test the umtxq_check_susp() change. Do I just need to ptrace the process? --=20 Regards, Bryan Drewery --p9Fqr5lLFAnCGEL5t8G4mRlXehaxggNpR-- --hwljxk2k8uM2tHhElvSoAgtwMvoOhwLgC Content-Type: application/pgp-signature; name="signature.asc" Content-Description: OpenPGP digital signature Content-Disposition: attachment; filename="signature.asc" -----BEGIN PGP SIGNATURE----- iQEzBAEBCAAdFiEE+Rc8ssOq6npcih8JNddxu25Gl88FAl1BDa0ACgkQNddxu25G l88UyAgAogITg3i08Cm7qOU4PKiomYEzSFl4ErRIe96narAXCBgPEA4V2/M9uiQv 5vVC37MB+BuBw8dbGASgnbaTEcL8GRxbt+t98fIoaTMC4BtqR6+gWPlKVn+T2YTa wkeLcfXI4yqW2haD4J5WFFfhUDgqeT79H1arufggpHnUM/tE8LjPujIQrSFAA3A5 ZFCyvFa7FxgFCNn1R3UDgbQ3dLnqPBn6mISfJyneYuo8IS5tZhMZI+0XuEz0lAnC /h4c4/Q11xyOqm6hZX07YcfzKVIyGBHTDrpKeykBIqLSMXqOfx0wNDR72ZhLIDny 0RMIja7p0YpULDsLhR5eFSuAyNLGVg== =hLC5 -----END PGP SIGNATURE----- --hwljxk2k8uM2tHhElvSoAgtwMvoOhwLgC--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5d3ffe05-58f3-ea26-13ed-51cdcd1deeaa>