Page MenuHomeFreeBSD
Differential D22688

multimedia/x264 - use base compiler on PowerPC64/ELFv2
ClosedPublic
Actions

Authored by alfredo on Dec 5 2019, 12:50 PM.
Tags
Referenced Files
Unknown Object (File)
Mon, Apr 7, 4:53 AM
Unknown Object (File)
Thu, Apr 3, 4:45 PM
Unknown Object (File)
Mar 3 2025, 1:00 AM
Unknown Object (File)
Mar 1 2025, 8:47 PM
Unknown Object (File)
Feb 26 2025, 10:19 AM
Unknown Object (File)
Feb 22 2025, 6:31 AM
Unknown Object (File)
Feb 22 2025, 3:37 AM
Unknown Object (File)
Feb 4 2025, 10:12 AM
View All Files
Subscribers
bdragon
koobs
luporl

Details

Reviewers
koobs
linimon
pkubaj
mikael
Commits
rP520083: multimedia/x264 - use base compiler on powerpc* when using clang
Summary

Environment running PowerPC64 with the ELFv2 uses llvm9 as base compiler, so we can remove GCC dependency.

Diff Detail

Repository
rP FreeBSD ports repository
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

alfredo created this revision.Dec 5 2019, 12:50 PM
Herald added subscribers: luporl, bdragon. · View Herald TranscriptDec 5 2019, 12:50 PM
Harbormaster completed remote builds in B27976: Diff 65260.Dec 5 2019, 12:50 PM
pkubaj requested changes to this revision.Dec 5 2019, 2:03 PM

powerpc and powerpcspe also switch to LLVM, so we should change default options there as well.

This revision now requires changes to proceed.Dec 5 2019, 2:03 PM
alfredo updated this revision to Diff 65547.Dec 12 2019, 3:13 PM

check compiler type

Harbormaster completed remote builds in B28102: Diff 65547.Dec 12 2019, 3:13 PM
alfredo updated this revision to Diff 65563.Dec 12 2019, 8:03 PM

remove explicit GCC dependency and add "modern compiler dependency" instead (compiler:c11)

Harbormaster completed remote builds in B28110: Diff 65563.Dec 12 2019, 8:03 PM
pkubaj requested changes to this revision.Dec 12 2019, 8:52 PM

Now it's not necessary to put include bsd.port.pre.mk and bsd.port.post.mk separately.

This revision now requires changes to proceed.Dec 12 2019, 8:52 PM
pkubaj added a comment.Dec 12 2019, 8:53 PM

Also, PGO options seems to require GCC. Why do you remove USE_GCC=any there? Have you verified it's not needed?

alfredo added a comment.Dec 13 2019, 5:23 PM
In D22688#498747, @pkubaj wrote:

Also, PGO options seems to require GCC. Why do you remove USE_GCC=any there? Have you verified it's not needed?

I'll revert this part to keep safe, affects behavior on platforms other than powerpc*

alfredo updated this revision to Diff 65598.Dec 13 2019, 5:27 PM

reverted PGO part

Harbormaster completed remote builds in B28127: Diff 65598.Dec 13 2019, 5:27 PM
koobs added a comment.Dec 14 2019, 12:19 AM

Can we get a proposed commit log message in this reviews summary that explains the context/issue and the 'why' of the change please

@pkubaj I'm happy for you to take this. Would you mind taking this to resolution, pending appropriate QA (poudriere, etc)?

pkubaj added a comment.Dec 14 2019, 1:19 AM
In D22688#499365, @koobs wrote:

Can we get a proposed commit log message in this reviews summary that explains the context/issue and the 'why' of the change please

@pkubaj I'm happy for you to take this. Would you mind taking this to resolution, pending appropriate QA (poudriere, etc)?

No problem, could you then resign from being a reviewer here?

pkubaj accepted this revision.Dec 14 2019, 1:21 AM

Still includes unnecessarily bsd.port.pre.mk. but I will fix it in the commit.

This revision is now accepted and ready to land.Dec 14 2019, 1:21 AM
koobs added a comment.Dec 14 2019, 1:34 AM

@pkubaj I believe you can "Commandeer Revision"

koobs added a subscriber: koobs.Dec 14 2019, 1:34 AM
In D22688#499387, @koobs wrote:

@pkubaj I believe you can "Commandeer Revision"

Not that you need to.

Approved by: koobs (maintainer)

Pending QA

Closed by commit rP520083: multimedia/x264 - use base compiler on powerpc* when using clang (authored by pkubaj). · Explain WhyDec 14 2019, 1:47 AM
This revision was automatically updated to reflect the committed changes.
pkubaj added a commit: rP520083: multimedia/x264 - use base compiler on powerpc* when using clang.

Revision Contents
Changeset List

PathSize
head/
multimedia/
x264/
9 lines

Diff 65643

View Options

head/multimedia/x264/Makefile

Show First 20 Lines • Show All 426 Lines • ▼ Show 20 Lines
SYSCTL_NODE(_net_link, OID_AUTO, vxlan, CTLFLAG_RW | CTLFLAG_MPSAFE, 0, SYSCTL_NODE(_net_link, OID_AUTO, vxlan, CTLFLAG_RW | CTLFLAG_MPSAFE, 0,
"Virtual eXtensible Local Area Network"); "Virtual eXtensible Local Area Network");
static int vxlan_legacy_port = 0; static int vxlan_legacy_port = 0;
TUNABLE_INT("net.link.vxlan.legacy_port", &vxlan_legacy_port); TUNABLE_INT("net.link.vxlan.legacy_port", &vxlan_legacy_port);
static int vxlan_reuse_port = 0; static int vxlan_reuse_port = 0;
TUNABLE_INT("net.link.vxlan.reuse_port", &vxlan_reuse_port); TUNABLE_INT("net.link.vxlan.reuse_port", &vxlan_reuse_port);
/*
* This macro controls the default upper limitation on nesting of vxlan
* tunnels. By default it is 3, as the overhead of IPv6 vxlan tunnel is 70
* bytes, this will create at most 210 bytes overhead and the most inner
* tunnel's MTU will be 1290 which will meet IPv6 minimum MTU size 1280.
* Be careful to configure the tunnels when raising the limit. A large
kpUnsubmitted
Not Done
Inline Actions

It's not quite that risky, because if_tunnel_check_nesting() checks if the last tunnel hop was this struct ifnet, so if users configure a loop the packet will be discarded immediately, not when it hits the nesting limit.

We do still need the limit, because users might configure a large number of different nested tunnels, which could still run us out of stack space, but that's harder to do by accident.

kp: It's not quite that risky, because if_tunnel_check_nesting() checks if the last tunnel hop was…
zleiAuthorUnsubmitted
Done
Inline Actions

It's not quite that risky, because if_tunnel_check_nesting() checks if the last tunnel hop was this struct ifnet, so if users configure a loop the packet will be discarded immediately, not when it hits the nesting limit.

I rechecked implementation of if_tunnel_check_nesting() and you are right :) Thanks for catching that.

We do still need the limit, because users might configure a large number of different nested tunnels, which could still run us out of stack space, but that's harder to do by accident.

True indeed.

zlei: > It's not quite that risky, because if_tunnel_check_nesting() checks if the last tunnel hop…
* number of nested tunnels can introduce system crash.
*/
#ifndef MAX_VXLAN_NEST
#define MAX_VXLAN_NEST 3
#endif
static int max_vxlan_nesting = MAX_VXLAN_NEST;
SYSCTL_INT(_net_link_vxlan, OID_AUTO, max_nesting, CTLFLAG_RW,
&max_vxlan_nesting, 0, "Max nested tunnels");
/* Default maximum number of addresses in the forwarding table. */ /* Default maximum number of addresses in the forwarding table. */
#ifndef VXLAN_FTABLE_MAX #ifndef VXLAN_FTABLE_MAX
#define VXLAN_FTABLE_MAX 2000 #define VXLAN_FTABLE_MAX 2000
#endif #endif
/* Timeout (in seconds) of addresses learned in the forwarding table. */ /* Timeout (in seconds) of addresses learned in the forwarding table. */
#ifndef VXLAN_FTABLE_TIMEOUT #ifndef VXLAN_FTABLE_TIMEOUT
#define VXLAN_FTABLE_TIMEOUT (20 * 60) #define VXLAN_FTABLE_TIMEOUT (20 * 60)
▲ Show 20 Lines • Show All 2,273 Lines • ▼ Show 20 Lines#ifdef INET6
return (error); return (error);
#else #else
m_freem(m); m_freem(m);
return (ENOTSUP); return (ENOTSUP);
#endif #endif
} }
#define MTAG_VXLAN_LOOP 0x7876706c /* vxlp */
static int static int
vxlan_transmit(struct ifnet *ifp, struct mbuf *m) vxlan_transmit(struct ifnet *ifp, struct mbuf *m)
{ {
struct rm_priotracker tracker; struct rm_priotracker tracker;
union vxlan_sockaddr vxlsa; union vxlan_sockaddr vxlsa;
struct vxlan_softc *sc; struct vxlan_softc *sc;
struct vxlan_ftable_entry *fe; struct vxlan_ftable_entry *fe;
struct ifnet *mcifp; struct ifnet *mcifp;
struct ether_header *eh; struct ether_header *eh;
int ipv4, error; int ipv4, error;
sc = ifp->if_softc; sc = ifp->if_softc;
eh = mtod(m, struct ether_header *); eh = mtod(m, struct ether_header *);
fe = NULL; fe = NULL;
mcifp = NULL; mcifp = NULL;
ETHER_BPF_MTAP(ifp, m); ETHER_BPF_MTAP(ifp, m);
VXLAN_RLOCK(sc, &tracker); VXLAN_RLOCK(sc, &tracker);
M_SETFIB(m, sc->vxl_fibnum); M_SETFIB(m, sc->vxl_fibnum);
if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) { if ((ifp->if_drv_flags & IFF_DRV_RUNNING) == 0) {
VXLAN_RUNLOCK(sc, &tracker); VXLAN_RUNLOCK(sc, &tracker);
m_freem(m); m_freem(m);
return (ENETDOWN); return (ENETDOWN);
}
if (__predict_false(if_tunnel_check_nesting(ifp, m, MTAG_VXLAN_LOOP,
max_vxlan_nesting) != 0)) {
VXLAN_RUNLOCK(sc, &tracker);
m_freem(m);
if_inc_counter(ifp, IFCOUNTER_OERRORS, 1);
return (ELOOP);
} }
if ((m->m_flags & (M_BCAST | M_MCAST)) == 0) if ((m->m_flags & (M_BCAST | M_MCAST)) == 0)
fe = vxlan_ftable_entry_lookup(sc, eh->ether_dhost); fe = vxlan_ftable_entry_lookup(sc, eh->ether_dhost);
if (fe == NULL) if (fe == NULL)
fe = &sc->vxl_default_fe; fe = &sc->vxl_default_fe;
vxlan_sockaddr_copy(&vxlsa, &fe->vxlfe_raddr.sa); vxlan_sockaddr_copy(&vxlsa, &fe->vxlfe_raddr.sa);
▲ Show 20 Lines • Show All 944 Lines • Show Last 20 Lines