- Linux-stable-mirror - lists.linaro.org

[PATCH 27/28] perf intel-pt: Insert callchain context into synthesized callchains

by Arnaldo Carvalho de Melo

From: Adrian Hunter <adrian.hunter(a)intel.com> In the absence of a fallback, callchains must encode also the callchain context. Do that now there is no fallback. Signed-off-by: Adrian Hunter <adrian.hunter(a)intel.com> Reviewed-by: Jiri Olsa <jolsa(a)kernel.org> Cc: Andi Kleen <ak(a)linux.intel.com> Cc: David S. Miller <davem(a)davemloft.net> Cc: Leo Yan <leo.yan(a)linaro.org> Cc: Mathieu Poirier <mathieu.poirier(a)linaro.org> Cc: stable(a)vger.kernel.org # 4.19 Link: http://lkml.kernel.org/r/100ea2ec-ed14-b56d-d810-e0a6d2f4b069@intel.com Signed-off-by: Arnaldo Carvalho de Melo <acme(a)redhat.com> --- tools/perf/util/intel-pt.c | 6 ++++-- tools/perf/util/thread-stack.c | 44 +++++++++++++++++++++++++++++++++--------- tools/perf/util/thread-stack.h | 2 +- 3 files changed, 40 insertions(+), 12 deletions(-) diff --git a/tools/perf/util/intel-pt.c b/tools/perf/util/intel-pt.c index ffa385a029b3..60732213d16a 100644 --- a/tools/perf/util/intel-pt.c +++ b/tools/perf/util/intel-pt.c @@ -759,7 +759,8 @@ static struct intel_pt_queue *intel_pt_alloc_queue(struct intel_pt *pt, if (pt->synth_opts.callchain) { size_t sz = sizeof(struct ip_callchain); - sz += pt->synth_opts.callchain_sz * sizeof(u64); + /* Add 1 to callchain_sz for callchain context */ + sz += (pt->synth_opts.callchain_sz + 1) * sizeof(u64); ptq->chain = zalloc(sz); if (!ptq->chain) goto out_free; @@ -1160,7 +1161,8 @@ static void intel_pt_prep_sample(struct intel_pt *pt, if (pt->synth_opts.callchain) { thread_stack__sample(ptq->thread, ptq->chain, - pt->synth_opts.callchain_sz, sample->ip); + pt->synth_opts.callchain_sz + 1, + sample->ip, pt->kernel_start); sample->callchain = ptq->chain; } diff --git a/tools/perf/util/thread-stack.c b/tools/perf/util/thread-stack.c index c091635bf7dc..61a4286a74dc 100644 --- a/tools/perf/util/thread-stack.c +++ b/tools/perf/util/thread-stack.c @@ -310,20 +310,46 @@ void thread_stack__free(struct thread *thread) } } +static inline u64 callchain_context(u64 ip, u64 kernel_start) +{ + return ip < kernel_start ? PERF_CONTEXT_USER : PERF_CONTEXT_KERNEL; +} + void thread_stack__sample(struct thread *thread, struct ip_callchain *chain, - size_t sz, u64 ip) + size_t sz, u64 ip, u64 kernel_start) { - size_t i; + u64 context = callchain_context(ip, kernel_start); + u64 last_context; + size_t i, j; - if (!thread || !thread->ts) - chain->nr = 1; - else - chain->nr = min(sz, thread->ts->cnt + 1); + if (sz < 2) { + chain->nr = 0; + return; + } - chain->ips[0] = ip; + chain->ips[0] = context; + chain->ips[1] = ip; + + if (!thread || !thread->ts) { + chain->nr = 2; + return; + } + + last_context = context; + + for (i = 2, j = 1; i < sz && j <= thread->ts->cnt; i++, j++) { + ip = thread->ts->stack[thread->ts->cnt - j].ret_addr; + context = callchain_context(ip, kernel_start); + if (context != last_context) { + if (i >= sz - 1) + break; + chain->ips[i++] = context; + last_context = context; + } + chain->ips[i] = ip; + } - for (i = 1; i < chain->nr; i++) - chain->ips[i] = thread->ts->stack[thread->ts->cnt - i].ret_addr; + chain->nr = i; } struct call_return_processor * diff --git a/tools/perf/util/thread-stack.h b/tools/perf/util/thread-stack.h index b7e41c4ebfdd..f97c00a8c251 100644 --- a/tools/perf/util/thread-stack.h +++ b/tools/perf/util/thread-stack.h @@ -84,7 +84,7 @@ int thread_stack__event(struct thread *thread, u32 flags, u64 from_ip, u64 to_ip, u16 insn_len, u64 trace_nr); void thread_stack__set_trace_nr(struct thread *thread, u64 trace_nr); void thread_stack__sample(struct thread *thread, struct ip_callchain *chain, - size_t sz, u64 ip); + size_t sz, u64 ip, u64 kernel_start); int thread_stack__flush(struct thread *thread); void thread_stack__free(struct thread *thread); size_t thread_stack__depth(struct thread *thread); -- 2.14.4

7 years, 1 month

1
0
0 0

[PATCH 14/28] perf callchain: Honour the ordering of PERF_CONTEXT_{USER, KERNEL, etc}

by Arnaldo Carvalho de Melo

From: "David S. Miller" <davem(a)davemloft.net> When processing using 'perf report -g caller', which is the default, we ended up reverting the callchain entries received from the kernel, but simply reverting throws away the information that tells that from a point onwards the addresses are for userspace, kernel, guest kernel, guest user, hypervisor. The idea is that if we are walking backwards, for each cluster of non-cpumode entries we have to first scan backwards for the next one and use that for the cluster. This seems silly and more expensive than it needs to be but it is enough for a initial fix. The code here is really complicated because it is intimately intertwined with the lbr and branch handling, as well as this callchain order, further fixes will be needed to properly take into account the cpumode in those cases. Another problem with ORDER_CALLER is that the NULL "0" IP that is at the end of most callchains shows up at the top of the histogram because every callchain contains it and with ORDER_CALLER it is the first entry. Signed-off-by: David S. Miller <davem(a)davemloft.net> Tested-by: Arnaldo Carvalho de Melo <acme(a)redhat.com> Cc: Adrian Hunter <adrian.hunter(a)intel.com> Cc: David Ahern <dsahern(a)gmail.com> Cc: Jiri Olsa <jolsa(a)kernel.org> Cc: Namhyung Kim <namhyung(a)kernel.org> Cc: Souvik Banerjee <souvik1997(a)gmail.com> Cc: Wang Nan <wangnan0(a)huawei.com> Cc: stable(a)vger.kernel.org # 4.19 Link: https://lkml.kernel.org/n/tip-2wt3ayp6j2y2f2xowixa8y6y@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo <acme(a)redhat.com> --- tools/perf/util/machine.c | 35 ++++++++++++++++++++++++++++++++++- 1 file changed, 34 insertions(+), 1 deletion(-) diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c index 111ae858cbcb..8ee8ab39d8ac 100644 --- a/tools/perf/util/machine.c +++ b/tools/perf/util/machine.c @@ -2140,6 +2140,27 @@ static int resolve_lbr_callchain_sample(struct thread *thread, return 0; } +static int find_prev_cpumode(struct ip_callchain *chain, struct thread *thread, + struct callchain_cursor *cursor, + struct symbol **parent, + struct addr_location *root_al, + u8 *cpumode, int ent) +{ + int err = 0; + + while (--ent >= 0) { + u64 ip = chain->ips[ent]; + + if (ip >= PERF_CONTEXT_MAX) { + err = add_callchain_ip(thread, cursor, parent, + root_al, cpumode, ip, + false, NULL, NULL, 0); + break; + } + } + return err; +} + static int thread__resolve_callchain_sample(struct thread *thread, struct callchain_cursor *cursor, struct perf_evsel *evsel, @@ -2246,6 +2267,12 @@ static int thread__resolve_callchain_sample(struct thread *thread, } check_calls: + if (callchain_param.order != ORDER_CALLEE) { + err = find_prev_cpumode(chain, thread, cursor, parent, root_al, + &cpumode, chain->nr - first_call); + if (err) + return (err < 0) ? err : 0; + } for (i = first_call, nr_entries = 0; i < chain_nr && nr_entries < max_stack; i++) { u64 ip; @@ -2260,9 +2287,15 @@ static int thread__resolve_callchain_sample(struct thread *thread, continue; #endif ip = chain->ips[j]; - if (ip < PERF_CONTEXT_MAX) ++nr_entries; + else if (callchain_param.order != ORDER_CALLEE) { + err = find_prev_cpumode(chain, thread, cursor, parent, + root_al, &cpumode, j); + if (err) + return (err < 0) ? err : 0; + continue; + } err = add_callchain_ip(thread, cursor, parent, root_al, &cpumode, ip, -- 2.14.4

7 years, 1 month

1
0
0 0

[PATCH v4] devres: Align data[] to ARCH_KMALLOC_MINALIGN

by Alexey Brodkin

Initially we bumped into problem with 32-bit aligned atomic64_t on ARC, see [1]. And then during quite lengthly discussion Peter Z. mentioned ARCH_KMALLOC_MINALIGN which IMHO makes perfect sense. If allocation is done by plain kmalloc() obtained buffer will be ARCH_KMALLOC_MINALIGN aligned and then why buffer obtained via devm_kmalloc() should have any other alignment? This way we at least get the same behavior for both types of allocation. [1] http://lists.infradead.org/pipermail/linux-snps-arc/2018-July/004009.html [2] http://lists.infradead.org/pipermail/linux-snps-arc/2018-July/004036.html Signed-off-by: Alexey Brodkin <abrodkin(a)synopsys.com> Cc: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Cc: Geert Uytterhoeven <geert(a)linux-m68k.org> Cc: David Laight <David.Laight(a)ACULAB.COM> Cc: Peter Zijlstra <peterz(a)infradead.org> Cc: Thomas Gleixner <tglx(a)linutronix.de> Cc: Vineet Gupta <vgupta(a)synopsys.com> Cc: Will Deacon <will.deacon(a)arm.com> Cc: Greg KH <greg(a)kroah.com> Cc: <stable(a)vger.kernel.org> # 4.8+ --- Changes v3 -> v4: * Use ARCH_KMALLOC_MINALIGN for alignment instead of "8" [Peter] Changes v2 -> v3: * Align explicitly to 8 bytes [David] * Rephrased in-line comment [David] * Added more techinical details to commit message [Greg] * Mention more alignment options in commit message [Geert] Changes v1 -> v2: * Reworded commit message * Inserted comment right in source [Thomas] drivers/base/devres.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/drivers/base/devres.c b/drivers/base/devres.c index 4aaf00d2098b..e038e2b3b7ea 100644 --- a/drivers/base/devres.c +++ b/drivers/base/devres.c @@ -26,8 +26,14 @@ struct devres_node { struct devres { struct devres_node node; - /* -- 3 pointers */ - unsigned long long data[]; /* guarantee ull alignment */ + /* + * Some archs want to perform DMA into kmalloc caches + * and need a guaranteed alignment larger than + * the alignment of a 64-bit integer. + * Thus we use ARCH_KMALLOC_MINALIGN here and get exactly the same + * buffer alignment as if it was allocated by plain kmalloc(). + */ + u8 __aligned(ARCH_KMALLOC_MINALIGN) data[]; }; struct devres_group { -- 2.17.2

7 years, 1 month

1
0
0 0

add da15fc2fa9c "perf tools: Disable parallelism for 'make clean'" to -stable

by Rasmus Villemoes

Hi Greg and other -stable maintainers Please consider adding da15fc2fa9c (perf tools: Disable parallelism for 'make clean') to your -stable trees. Spurious build failures like that make it harder to do automatic tests of new -stable (and -stable-rc) versions. Thanks, Rasmus

7 years, 1 month

2
1
0 0

stable regression: revert request for netfilter ipv6 defrag bug

by Florian Westphal

Hi, please consider reverting commit 84379c9afe011020e797e3f50a662b08a6355dcf netfilter: ipv6: nf_defrag: drop skb dst before queueing It causes kernel crash for locally generated ipv6 fragments when netfilter ipv6 defragmentation is used. The faulty commit is not essential for -stable, it only delays netns teardown for longer than needed when that netns still has ipv6 frags queued. Much better than crash :-/ commit ids are: 4.4.y: not affected (not backported) 4.9.y: backported as ad8b1ffc3efae2f65080bdb11145c87d299b8f9a 4.14.y: backported as 28c74ff85efd192aeca9005499ca50c24d795f61 4.18.y: (first affected kernel): 84379c9afe011020e797e3f50a662b08a6355dcf For 4.19.y, you could also wait for a bug fix to hit Linus tree, I can ping you again once its in: https://patchwork.ozlabs.org/patch/988233/ Thanks, Florian

7 years, 1 month

2
1
0 0

[PATCH] kdb: Fix btc (backtrace on each cpu) crash

by Douglas Anderson

Typing 'btc' on kdb doing all sorts of fail. Sometimes it would crash, sometimes display nothing, and sometimes hang. Bisect tracked this down to the commit ad67b74d2469 ("printk: hash addresses printed with %p"), suggesting an obvious fix. The pointer used internally in kdb shouldn't be hashed, so switch it to %px. Fixes: ad67b74d2469 ("printk: hash addresses printed with %p") Cc: stable(a)vger.kernel.org Signed-off-by: Douglas Anderson <dianders(a)chromium.org> --- kernel/debug/kdb/kdb_bt.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/kernel/debug/kdb/kdb_bt.c b/kernel/debug/kdb/kdb_bt.c index 6ad4a9fcbd6f..7921ae4fca8d 100644 --- a/kernel/debug/kdb/kdb_bt.c +++ b/kernel/debug/kdb/kdb_bt.c @@ -179,14 +179,14 @@ kdb_bt(int argc, const char **argv) kdb_printf("no process for cpu %ld\n", cpu); return 0; } - sprintf(buf, "btt 0x%p\n", KDB_TSK(cpu)); + sprintf(buf, "btt 0x%px\n", KDB_TSK(cpu)); kdb_parse(buf); return 0; } kdb_printf("btc: cpu status: "); kdb_parse("cpu\n"); for_each_online_cpu(cpu) { - sprintf(buf, "btt 0x%p\n", KDB_TSK(cpu)); + sprintf(buf, "btt 0x%px\n", KDB_TSK(cpu)); kdb_parse(buf); touch_nmi_watchdog(); } -- 2.19.1.568.g152ad8e336-goog

7 years, 1 month

3
2
0 0

request for 4.14-stable: f9a1c372299f ("ALSA: usx2y: Fix invalid stream URBs")

by Sudip Mukherjee

Hi Greg, This was not marked for stable but seems it should be in stable. Please apply to your queue of 4.14-stable. -- Regards Sudip

7 years, 1 month

2
1
0 0

request for 4.14-stable: f9ffcb0a21e1 ("media: uvcvideo: Fix driver reference counting")

by Sudip Mukherjee

Hi Greg, This was not marked for stable but seems it should be in stable. Please apply to your queue of 4.14-stable. -- Regards Sudip

7 years, 1 month

2
1
0 0

request for 4.14-stable: 5663d8f9bbe4 ("kvm: x86: fix WARN due to uninitialized guest FPU state")

by Sudip Mukherjee

Hi Greg, This was missing in 4.14-stable. Please apply to your queue. -- Regards Sudip

7 years, 1 month

3
2
0 0

[4.4/4.9/4.14 patch] mtd: spi-nor: Add support for is25wp series chips

by Adrian Bunk

From: Kimmo Rautkoski <ext-kimmo.rautkoski(a)vaisala.com> Added support for is25wp032, is25wp064 and is25wp128. Signed-off-by: Kimmo Rautkoski <ext-kimmo.rautkoski(a)vaisala.com> Reviewed-by: Marek Vasut <marek.vasut(a)gmail.com> Signed-off-by: Boris Brezillon <boris.brezillon(a)bootlin.com> [ Adrian Bunk: Trivial adaption to changed context. ] Signed-off-by: Adrian Bunk <bunk(a)kernel.org> --- drivers/mtd/spi-nor/spi-nor.c | 6 ++++++ 1 file changed, 6 insertions(+) diff --git a/drivers/mtd/spi-nor/spi-nor.c b/drivers/mtd/spi-nor/spi-nor.c index 19c000722cbc..34ecc12ee3d9 100644 --- a/drivers/mtd/spi-nor/spi-nor.c +++ b/drivers/mtd/spi-nor/spi-nor.c @@ -1005,6 +1005,12 @@ static const struct flash_info spi_nor_ids[] = { /* ISSI */ { "is25cd512", INFO(0x7f9d20, 0, 32 * 1024, 2, SECT_4K) }, + { "is25wp032", INFO(0x9d7016, 0, 64 * 1024, 64, + SECT_4K | SPI_NOR_DUAL_READ | SPI_NOR_QUAD_READ) }, + { "is25wp064", INFO(0x9d7017, 0, 64 * 1024, 128, + SECT_4K | SPI_NOR_DUAL_READ | SPI_NOR_QUAD_READ) }, + { "is25wp128", INFO(0x9d7018, 0, 64 * 1024, 256, + SECT_4K | SPI_NOR_DUAL_READ | SPI_NOR_QUAD_READ) }, /* Macronix */ { "mx25l512e", INFO(0xc22010, 0, 64 * 1024, 1, SECT_4K) }, -- 2.11.0

7 years, 1 month

3
3
0 0

[PATCH 4.18] Revert "mm: slowly shrink slabs with a relatively small number of objects"

by Sasha Levin

This reverts commit 62aad93f09c1952ede86405894df1b22012fd5ab. Which was upstream commit 172b06c32b94 ("mm: slowly shrink slabs with a relatively small number of objects"). The upstream commit was found to cause regressions. While there is a proposed fix upstream, revent this patch from stable trees for now as testing the fix will take some time. Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- mm/vmscan.c | 11 ----------- 1 file changed, 11 deletions(-) diff --git a/mm/vmscan.c b/mm/vmscan.c index fc0436407471..03822f86f288 100644 --- a/mm/vmscan.c +++ b/mm/vmscan.c @@ -386,17 +386,6 @@ static unsigned long do_shrink_slab(struct shrink_control *shrinkctl, delta = freeable >> priority; delta *= 4; do_div(delta, shrinker->seeks); - - /* - * Make sure we apply some minimal pressure on default priority - * even on small cgroups. Stale objects are not only consuming memory - * by themselves, but can also hold a reference to a dying cgroup, - * preventing it from being reclaimed. A dying cgroup with all - * corresponding structures like per-cpu stats and kmem caches - * can be really big, so it may lead to a significant waste of memory. - */ - delta = max_t(unsigned long long, delta, min(freeable, batch_size)); - total_scan += delta; if (total_scan < 0) { pr_err("shrink_slab: %pF negative objects to delete nr=%ld\n", -- 2.17.1

7 years, 1 month

1
1
0 0

[PATCH v4.14-stable] sch_netem: restore skb->dev after dequeuing from the rbtree

by Christoph Paasch

Upstream commit bffa72cf7f9d ("net: sk_buff rbnode reorg") got backported as commit 6b921536f170 ("net: sk_buff rbnode reorg") into the v4.14.x-tree. However, the backport does not include the changes in sch_netem.c We need these, as otherwise the skb->dev pointer is not set when dequeueing from the netem rbtree, resulting in a panic: [ 15.427748] BUG: unable to handle kernel NULL pointer dereference at 00000000000000d0 [ 15.428863] IP: netif_skb_features+0x24/0x230 [ 15.429402] PGD 0 P4D 0 [ 15.429733] Oops: 0000 [#1] SMP PTI [ 15.430169] Modules linked in: [ 15.430614] CPU: 3 PID: 0 Comm: swapper/3 Not tainted 4.14.77.mptcp #77 [ 15.431497] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 0.5.1 01/01/2011 [ 15.432568] task: ffff88042db19680 task.stack: ffffc90000070000 [ 15.433356] RIP: 0010:netif_skb_features+0x24/0x230 [ 15.433977] RSP: 0018:ffff88043fd83e70 EFLAGS: 00010286 [ 15.434665] RAX: ffff880429ad80c0 RBX: ffff88042bd0e400 RCX: ffff880429ad8000 [ 15.435585] RDX: 0000000000000000 RSI: 0000000000000000 RDI: ffff88042bd0e400 [ 15.436551] RBP: ffff88042bd0e400 R08: ffff88042a4b6c9c R09: 0000000000000001 [ 15.437485] R10: 0000000000000004 R11: 0000000000000000 R12: ffff88042c700000 [ 15.438393] R13: ffff88042c700000 R14: ffff88042a4b6c00 R15: ffff88042c6bb000 [ 15.439315] FS: 0000000000000000(0000) GS:ffff88043fd80000(0000) knlGS:0000000000000000 [ 15.440314] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 15.441084] CR2: 00000000000000d0 CR3: 000000042c374000 CR4: 00000000000006e0 [ 15.442016] Call Trace: [ 15.442333] <IRQ> [ 15.442596] validate_xmit_skb+0x17/0x270 [ 15.443134] validate_xmit_skb_list+0x38/0x60 [ 15.443698] sch_direct_xmit+0x102/0x190 [ 15.444198] __qdisc_run+0xe3/0x240 [ 15.444671] net_tx_action+0x121/0x140 [ 15.445177] __do_softirq+0xe2/0x224 [ 15.445654] irq_exit+0xbf/0xd0 [ 15.446072] smp_apic_timer_interrupt+0x5d/0x90 [ 15.446654] apic_timer_interrupt+0x7d/0x90 [ 15.447185] </IRQ> [ 15.447460] RIP: 0010:native_safe_halt+0x2/0x10 [ 15.447992] RSP: 0018:ffffc90000073f10 EFLAGS: 00000282 ORIG_RAX: ffffffffffffff10 [ 15.449008] RAX: ffffffff816667d0 RBX: ffffffff820946b0 RCX: 0000000000000000 [ 15.449895] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000 [ 15.450768] RBP: ffffffff82026940 R08: 00000004e858e5e1 R09: ffff88042a4b6d58 [ 15.451643] R10: 0000000000000000 R11: 000000d0d56879bb R12: 0000000000000000 [ 15.452478] R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000 [ 15.453340] ? __sched_text_end+0x2/0x2 [ 15.453835] default_idle+0xf/0x20 [ 15.454259] do_idle+0x170/0x200 [ 15.454653] cpu_startup_entry+0x14/0x20 [ 15.455142] secondary_startup_64+0xa5/0xb0 [ 15.455715] Code: 1f 84 00 00 00 00 00 55 53 48 89 fd 48 83 ec 08 8b 87 bc 00 00 00 48 8b 8f c0 00 00 00 0f b6 97 81 00 00 00 48 8b 77 10 48 01 c8 <48> 8b 9 [ 15.458138] RIP: netif_skb_features+0x24/0x230 RSP: ffff88043fd83e70 [ 15.458933] CR2: 00000000000000d0 [ 15.459352] ---[ end trace 083925903ae60570 ]--- Fixes: 6b921536f170 ("net: sk_buff rbnode reorg") Cc: Stephen Hemminger <stephen(a)networkplumber.org> Cc: Eric Dumazet <edumazet(a)google.com> Cc: Soheil Hassas Yeganeh <soheil(a)google.com> Cc: Wei Wang <weiwan(a)google.com> Cc: Willem de Bruijn <willemb(a)google.com> Signed-off-by: Christoph Paasch <cpaasch(a)apple.com> --- Notes: This patch should only make it into v4.14-stable as that's the only branch where the offending commit has been backported to. net/sched/sch_netem.c | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/net/sched/sch_netem.c b/net/sched/sch_netem.c index 2a2ab6bfe5d8..3d325b840802 100644 --- a/net/sched/sch_netem.c +++ b/net/sched/sch_netem.c @@ -624,6 +624,10 @@ static struct sk_buff *netem_dequeue(struct Qdisc *sch) skb->next = NULL; skb->prev = NULL; skb->tstamp = netem_skb_cb(skb)->tstamp_save; + /* skb->dev shares skb->rbnode area, + * we need to restore its value. + */ + skb->dev = qdisc_dev(sch); #ifdef CONFIG_NET_CLS_ACT /* -- 2.16.2

7 years, 1 month

4
3
0 0

Automated testing for stable kernel branches?

by Major Hayden

Hello there, I am working on a project at Red Hat where we do quick testing on patches for internal kernels before they merge. The goal is to catch bugs or issues before they merge into kernel trees and avoid situations where kernels need time-consuming bisects when lots of patches are merged at once. We aim to put valuable feedback into a kernel developer's inbox within four hours. Our team has built a pipeline where we merge patches, compile kernels (for various architectures), and run tests on real hardware (various architectures). The current test set is fairly basic and it includes LTP plus some additional open source tests. We are looking to gradually expand those over time as we evaluate which tests provide the most value and find the most problems. We would love to bring this to upstream kernel repositories and we thought that linux-stable might be a good place to start. The developer/maintainer experience would look something like this: 1) Developer submits a patchset 2) Those patches end up in Patchwork 3) We pull patches from patchwork, compile kernels, and test them 4) We reply to the thread on the mailing list with a brief set of results (one time per patchset) Developers do not need to change any existing workflows. We gather the patches, test them, and reply in the appropriate place. Is this something that the linux-stable community and maintainers would find valuable? If so, feel free to ask any questions about our process and we can go over any of those parts in more detail. If not, please let me know anyway! Our team is always looking for ways to improve. :) Thanks a bunch for reading this far and I look forward to hearing from you. -- Major Hayden

7 years, 1 month

3
7
0 0

Re: Regression found (Stop-marking-clocks-as-CLK_IS_CRITICAL)

by Hans de Goede

Hi, On 31-10-18 07:02, Mogens Jensen wrote: > ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ > On Tuesday, October 30, 2018 7:10 PM, Hans de Goede <hdegoede(a)redhat.com> wrote: > >> Hi, >> >> On 30-10-18 19:56, Mogens Jensen wrote: >> >>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ >>> On Tuesday, October 30, 2018 4:04 PM, Hans de Goede hdegoede(a)redhat.com wrote: >>> >>>> Hi, >>>> On 30-10-18 16:46, Hans de Goede wrote: >>>> >>>>> Hi, >>>>> On 30-10-18 16:04, Pierre-Louis Bossart wrote: >>>>> >>>>>> In addition I am not aware of any baytrail device using plt_clk_0, so moving a common machine driver such a cht_bsw_max98090_ti to use plt_clk0 only would break other devices (e.g. Rambi/Orco). Asking for both clocks to be on might work though, >>>>> >>>>> Ok, so we need to have a DMI based quirk for the Swanky and maybe also >>>>> the clapper to use plt_clk_0 there. Asking for 2 clks if we only need >>>>> one does not seem like a good plan. >>>> >>>> Dean, Mogens, >>>> To write a proper patch for this I'm going to need DMI strings >>>> from your devices. >>>> Can you please run (as normal user): >>>> grep . /sys/class/dmi/id/* 2> /dev/null >>>> And reply with the output of this command? >>>> I have attached the output from a coreboot seabios based clapper. >> >> Thank you. >> >>> Should I still test 0001-ASoC-intel-cht_bsw_max98090_ti-Use-pmc_plt_clk_0-ins.patch with SND_SOC_INTEL_CHT_BSW_MAX98090_TI_MACH and asoundrc from Dean? There seems to have been some development in the case since that request was made. >> >> Yes please test that, I expect that to also fix things for the >> Clapper, but I need to have that confirmed before submitting a >> patch upstream adding a quirk for the Clapper to use pmc_plt_clk_0 >> instead of pmc_plt_clk_3. >> >> Regards, >> >> Hans >> > Unfortunately I only have access to longterm kernel 4.14 for building/running on this system, and 0001-ASoC-intel-cht_bsw_max98090_ti-Use-pmc_plt_clk_0-ins.patch does not patch against 4.14.78. Can a test patch for 4.14 be created? Can you run (as root): for i in /sys/kernel/debug/clk/pmc_plt_clk_?; do echo -n "$i: "; cat $i/clk_flags; echo; done When running a kernel with working audio? Then I can confirm that the Clapper is also using pmc_plt_clk_0, so that I can fix this for the clapper for 4.18+ I've just checked the 4.14 sources and in 4.14 the SND_SOC_INTEL_CHT_BSW_MAX98090_TI_MACH driver does not support mclk control yet, so for the 4.14 kernel the only way to fix this is to revert the 648e921888ad ("clk: x86: Stop marking clocks as CLK_IS_CRITICAL") commit. Regards, Hans

7 years, 1 month

2
1
0 0

[PATCH] selftests/powerpc: Fix ptrace tm failure

by Breno Leitao

Test ptrace-tm-spd-gpr fails on current kernel (4.19) due to a segmentation fault that happens on the child process prior to setting cptr[2] = 1. This causes the parent process to wait forever at 'while (!pptr[2])' and the test to be killed by the test harness framework by timeout, thus, failing. The segmentation fault happens because of a inline assembly being generated as: 0x10000355c <tm_spd_gpr+492> lfs f0, 0(0) This is reading memory position 0x0 and causing the segmentation fault. This code is being generated by ASM_LOAD_FPR_SINGLE_PRECISION(flt_4), where flt_4 is passed to the inline assembly block as: [flt_4] "r" (&d) Since the inline assembly 'r' constraint means any GPR, gpr0 is being chosen, thus causing this issue when issuing a Load Floating-Point Single instruction. This patch simply changes the constraint to 'b', which specify that this register will be used as base, and r0 is not allowed to be used, avoiding this issue. Other than that, removing flt_2 register from the input operands, since it is not used by the inline assembly code at all. Cc: stable(a)vger.kernel.org Signed-off-by: Breno Leitao <leitao(a)debian.org> --- tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-gpr.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-gpr.c b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-gpr.c index 327fa943c7f3..dbdffa2e2c82 100644 --- a/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-gpr.c +++ b/tools/testing/selftests/powerpc/ptrace/ptrace-tm-spd-gpr.c @@ -67,8 +67,8 @@ void tm_spd_gpr(void) "3: ;" : [res] "=r" (result), [texasr] "=r" (texasr) : [gpr_1]"i"(GPR_1), [gpr_2]"i"(GPR_2), [gpr_4]"i"(GPR_4), - [sprn_texasr] "i" (SPRN_TEXASR), [flt_1] "r" (&a), - [flt_2] "r" (&b), [flt_4] "r" (&d) + [sprn_texasr] "i" (SPRN_TEXASR), [flt_1] "b" (&a), + [flt_4] "b" (&d) : "memory", "r5", "r6", "r7", "r8", "r9", "r10", "r11", "r12", "r13", "r14", "r15", "r16", "r17", "r18", "r19", "r20", "r21", "r22", "r23", -- 2.19.0

7 years, 1 month

3
2
0 0

[patch 55/96] kbuild: fix kernel/bounds.c 'W=1' warning

by akpm＠linux-foundation.org

From: Arnd Bergmann <arnd(a)arndb.de> Subject: kbuild: fix kernel/bounds.c 'W=1' warning Building any configuration with 'make W=1' produces a warning: kernel/bounds.c:16:6: warning: no previous prototype for 'foo' [-Wmissing-prototypes] When also passing -Werror, this prevents us from building any other files. Nobody ever calls the function, but we can't make it 'static' either since we want the compiler output. Calling it 'main' instead however avoids the warning, because gcc does not insist on having a declaration for main. Link: http://lkml.kernel.org/r/20181005083313.2088252-1-arnd@arndb.de Signed-off-by: Arnd Bergmann <arnd(a)arndb.de> Reported-by: Kieran Bingham <kieran.bingham+renesas(a)ideasonboard.com> Reviewed-by: Kieran Bingham <kieran.bingham+renesas(a)ideasonboard.com> Cc: David Laight <David.Laight(a)ACULAB.COM> Cc: Masahiro Yamada <yamada.masahiro(a)socionext.com> Cc: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- kernel/bounds.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) --- a/kernel/bounds.c~kbuild-fix-kernel-boundsc-w=1-warning +++ a/kernel/bounds.c @@ -13,7 +13,7 @@ #include <linux/log2.h> #include <linux/spinlock_types.h> -void foo(void) +int main(void) { /* The enum constants to put into include/generated/bounds.h */ DEFINE(NR_PAGEFLAGS, __NR_PAGEFLAGS); @@ -23,4 +23,6 @@ void foo(void) #endif DEFINE(SPINLOCK_SIZE, sizeof(spinlock_t)); /* End of constants */ + + return 0; } _

7 years, 1 month

1
0
0 0

[patch 03/96] mm/hmm: fix race between hmm_mirror_unregister() and mmu_notifier callback

by akpm＠linux-foundation.org

From: Ralph Campbell <rcampbell(a)nvidia.com> Subject: mm/hmm: fix race between hmm_mirror_unregister() and mmu_notifier callback In hmm_mirror_unregister(), mm->hmm is set to NULL and then mmu_notifier_unregister_no_release() is called. That creates a small window where mmu_notifier can call mmu_notifier_ops with mm->hmm equal to NULL. Fix this by first unregistering mmu notifier callbacks and then setting mm->hmm to NULL. Similarly in hmm_register(), set mm->hmm before registering mmu_notifier callbacks so callback functions always see mm->hmm set. Link: http://lkml.kernel.org/r/20181019160442.18723-4-jglisse@redhat.com Signed-off-by: Ralph Campbell <rcampbell(a)nvidia.com> Signed-off-by: Jérôme Glisse <jglisse(a)redhat.com> Reviewed-by: John Hubbard <jhubbard(a)nvidia.com> Reviewed-by: Jérôme Glisse <jglisse(a)redhat.com> Reviewed-by: Balbir Singh <bsingharora(a)gmail.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- mm/hmm.c | 36 +++++++++++++++++++++--------------- 1 file changed, 21 insertions(+), 15 deletions(-) --- a/mm/hmm.c~mm-hmm-fix-race-between-hmm_mirror_unregister-and-mmu_notifier-callback +++ a/mm/hmm.c @@ -91,16 +91,6 @@ static struct hmm *hmm_register(struct m spin_lock_init(&hmm->lock); hmm->mm = mm; - /* - * We should only get here if hold the mmap_sem in write mode ie on - * registration of first mirror through hmm_mirror_register() - */ - hmm->mmu_notifier.ops = &hmm_mmu_notifier_ops; - if (__mmu_notifier_register(&hmm->mmu_notifier, mm)) { - kfree(hmm); - return NULL; - } - spin_lock(&mm->page_table_lock); if (!mm->hmm) mm->hmm = hmm; @@ -108,12 +98,27 @@ static struct hmm *hmm_register(struct m cleanup = true; spin_unlock(&mm->page_table_lock); - if (cleanup) { - mmu_notifier_unregister(&hmm->mmu_notifier, mm); - kfree(hmm); - } + if (cleanup) + goto error; + + /* + * We should only get here if hold the mmap_sem in write mode ie on + * registration of first mirror through hmm_mirror_register() + */ + hmm->mmu_notifier.ops = &hmm_mmu_notifier_ops; + if (__mmu_notifier_register(&hmm->mmu_notifier, mm)) + goto error_mm; return mm->hmm; + +error_mm: + spin_lock(&mm->page_table_lock); + if (mm->hmm == hmm) + mm->hmm = NULL; + spin_unlock(&mm->page_table_lock); +error: + kfree(hmm); + return NULL; } void hmm_mm_destroy(struct mm_struct *mm) @@ -278,12 +283,13 @@ void hmm_mirror_unregister(struct hmm_mi if (!should_unregister || mm == NULL) return; + mmu_notifier_unregister_no_release(&hmm->mmu_notifier, mm); + spin_lock(&mm->page_table_lock); if (mm->hmm == hmm) mm->hmm = NULL; spin_unlock(&mm->page_table_lock); - mmu_notifier_unregister_no_release(&hmm->mmu_notifier, mm); kfree(hmm); } EXPORT_SYMBOL(hmm_mirror_unregister); _

7 years, 1 month

1
0
0 0

[patch 02/96] mm/rmap: map_pte() was not handling private ZONE_DEVICE page properly

by akpm＠linux-foundation.org

From: Ralph Campbell <rcampbell(a)nvidia.com> Subject: mm/rmap: map_pte() was not handling private ZONE_DEVICE page properly Private ZONE_DEVICE pages use a special pte entry and thus are not present. Properly handle this case in map_pte(), it is already handled in check_pte(), the map_pte() part was lost in some rebase most probably. Without this patch the slow migration path can not migrate back to any private ZONE_DEVICE memory to regular memory. This was found after stress testing migration back to system memory. This ultimatly can lead to the CPU constantly page fault looping on the special swap entry. Link: http://lkml.kernel.org/r/20181019160442.18723-3-jglisse@redhat.com Signed-off-by: Ralph Campbell <rcampbell(a)nvidia.com> Signed-off-by: Jérôme Glisse <jglisse(a)redhat.com> Reviewed-by: Balbir Singh <bsingharora(a)gmail.com> Cc: Andrew Morton <akpm(a)linux-foundation.org> Cc: Kirill A. Shutemov <kirill.shutemov(a)linux.intel.com> Cc: <stable(a)vger.kernel.org> Signed-off-by: Andrew Morton <akpm(a)linux-foundation.org> --- mm/page_vma_mapped.c | 24 +++++++++++++++++++++++- 1 file changed, 23 insertions(+), 1 deletion(-) --- a/mm/page_vma_mapped.c~mm-rmap-map_pte-was-not-handling-private-zone_device-page-properly-v3 +++ a/mm/page_vma_mapped.c @@ -21,7 +21,29 @@ static bool map_pte(struct page_vma_mapp if (!is_swap_pte(*pvmw->pte)) return false; } else { - if (!pte_present(*pvmw->pte)) + /* + * We get here when we are trying to unmap a private + * device page from the process address space. Such + * page is not CPU accessible and thus is mapped as + * a special swap entry, nonetheless it still does + * count as a valid regular mapping for the page (and + * is accounted as such in page maps count). + * + * So handle this special case as if it was a normal + * page mapping ie lock CPU page table and returns + * true. + * + * For more details on device private memory see HMM + * (include/linux/hmm.h or mm/hmm.c). + */ + if (is_swap_pte(*pvmw->pte)) { + swp_entry_t entry; + + /* Handle un-addressable ZONE_DEVICE memory */ + entry = pte_to_swp_entry(*pvmw->pte); + if (!is_device_private_entry(entry)) + return false; + } else if (!pte_present(*pvmw->pte)) return false; } } _

7 years, 1 month

1
0
0 0

Re: Regression found (Stop-marking-clocks-as-CLK_IS_CRITICAL)

by Hans de Goede

Hi, On 30-10-18 19:56, Mogens Jensen wrote: > ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐ > On Tuesday, October 30, 2018 4:04 PM, Hans de Goede <hdegoede(a)redhat.com> wrote: > >> Hi, >> >> On 30-10-18 16:46, Hans de Goede wrote: >> >>> Hi, >>> On 30-10-18 16:04, Pierre-Louis Bossart wrote: >>> >>>> In addition I am not aware of any baytrail device using plt_clk_0, so moving a common machine driver such a cht_bsw_max98090_ti to use plt_clk0 only would break other devices (e.g. Rambi/Orco). Asking for both clocks to be on might work though, >>> >>> Ok, so we need to have a DMI based quirk for the Swanky and maybe also >>> the clapper to use plt_clk_0 there. Asking for 2 clks if we only need >>> one does not seem like a good plan. >> >> Dean, Mogens, >> >> To write a proper patch for this I'm going to need DMI strings >> from your devices. >> >> Can you please run (as normal user): >> >> grep . /sys/class/dmi/id/* 2> /dev/null >> >> And reply with the output of this command? > I have attached the output from a coreboot seabios based clapper. Thank you. > Should I still test 0001-ASoC-intel-cht_bsw_max98090_ti-Use-pmc_plt_clk_0-ins.patch with SND_SOC_INTEL_CHT_BSW_MAX98090_TI_MACH and asoundrc from Dean? There seems to have been some development in the case since that request was made. Yes please test that, I expect that to also fix things for the Clapper, but I need to have that confirmed before submitting a patch upstream adding a quirk for the Clapper to use pmc_plt_clk_0 instead of pmc_plt_clk_3. Regards, Hans

7 years, 1 month

1
0
0 0

Re: [PATCH] soc: ti: QMSS: Fix usage of irq_set_affinity_hint

by Santosh Shilimkar

Hi Greg, On 10/30/2018 4:11 AM, Marc Zyngier wrote: > The Keystone QMSS driver is pretty damaged, in the sense that it > does things like this: > > irq_set_affinity_hint(irq, to_cpumask(&cpu_map)); > > where cpu_map is a local variable. As we leave the function, this > will point to nowhere-land, and things will end-up badly. > > Instead, let's use a proper cpumask that gets allocated, giving > the driver a chance to actually work with things like irqbalance > as well as have a hypothetical 64bit future. > > Signed-off-by: Marc Zyngier <marc.zyngier(a)arm.com> > --- > I found this one by inspection after finding a similar bug in an > unrelated driver. It is only compile-tested. It would probably > a Cc stable, but that's Santosh's decision. Would be able to apply this fix from Marc for stable or it needs to be re-posted with CC to stable ? > > drivers/soc/ti/knav_qmss.h | 4 ++-- > drivers/soc/ti/knav_qmss_acc.c | 10 +++++----- > drivers/soc/ti/knav_qmss_queue.c | 22 +++++++++++++++------- > 3 files changed, 22 insertions(+), 14 deletions(-) > > diff --git a/drivers/soc/ti/knav_qmss.h b/drivers/soc/ti/knav_qmss.h > index 3efc47e82973..bd040c29c4bf 100644 > --- a/drivers/soc/ti/knav_qmss.h > +++ b/drivers/soc/ti/knav_qmss.h > @@ -329,8 +329,8 @@ struct knav_range_ops { > }; > > struct knav_irq_info { > - int irq; > - u32 cpu_map; > + int irq; > + struct cpumask *cpu_mask; > }; > > struct knav_range_info { > diff --git a/drivers/soc/ti/knav_qmss_acc.c b/drivers/soc/ti/knav_qmss_acc.c > index 316e82e46f6c..2f7fb2dcc1d6 100644 > --- a/drivers/soc/ti/knav_qmss_acc.c > +++ b/drivers/soc/ti/knav_qmss_acc.c > @@ -205,18 +205,18 @@ static int knav_range_setup_acc_irq(struct knav_range_info *range, > { > struct knav_device *kdev = range->kdev; > struct knav_acc_channel *acc; > - unsigned long cpu_map; > + struct cpumask *cpu_mask; > int ret = 0, irq; > u32 old, new; > > if (range->flags & RANGE_MULTI_QUEUE) { > acc = range->acc; > irq = range->irqs[0].irq; > - cpu_map = range->irqs[0].cpu_map; > + cpu_mask = range->irqs[0].cpu_mask; > } else { > acc = range->acc + queue; > irq = range->irqs[queue].irq; > - cpu_map = range->irqs[queue].cpu_map; > + cpu_mask = range->irqs[queue].cpu_mask; > } > > old = acc->open_mask; > @@ -239,8 +239,8 @@ static int knav_range_setup_acc_irq(struct knav_range_info *range, > acc->name, acc->name); > ret = request_irq(irq, knav_acc_int_handler, 0, acc->name, > range); > - if (!ret && cpu_map) { > - ret = irq_set_affinity_hint(irq, to_cpumask(&cpu_map)); > + if (!ret && cpu_mask) { > + ret = irq_set_affinity_hint(irq, cpu_mask); > if (ret) { > dev_warn(range->kdev->dev, > "Failed to set IRQ affinity\n"); > diff --git a/drivers/soc/ti/knav_qmss_queue.c b/drivers/soc/ti/knav_qmss_queue.c > index b5d5673c255c..8b418379272d 100644 > --- a/drivers/soc/ti/knav_qmss_queue.c > +++ b/drivers/soc/ti/knav_qmss_queue.c > @@ -118,19 +118,17 @@ static int knav_queue_setup_irq(struct knav_range_info *range, > struct knav_queue_inst *inst) > { > unsigned queue = inst->id - range->queue_base; > - unsigned long cpu_map; > int ret = 0, irq; > > if (range->flags & RANGE_HAS_IRQ) { > irq = range->irqs[queue].irq; > - cpu_map = range->irqs[queue].cpu_map; > ret = request_irq(irq, knav_queue_int_handler, 0, > inst->irq_name, inst); > if (ret) > return ret; > disable_irq(irq); > - if (cpu_map) { > - ret = irq_set_affinity_hint(irq, to_cpumask(&cpu_map)); > + if (range->irqs[queue].cpu_mask) { > + ret = irq_set_affinity_hint(irq, range->irqs[queue].cpu_mask); > if (ret) { > dev_warn(range->kdev->dev, > "Failed to set IRQ affinity\n"); > @@ -1262,9 +1260,19 @@ static int knav_setup_queue_range(struct knav_device *kdev, > > range->num_irqs++; > > - if (IS_ENABLED(CONFIG_SMP) && oirq.args_count == 3) > - range->irqs[i].cpu_map = > - (oirq.args[2] & 0x0000ff00) >> 8; > + if (IS_ENABLED(CONFIG_SMP) && oirq.args_count == 3) { > + unsigned long mask; > + int bit; > + > + range->irqs[i].cpu_mask = devm_kzalloc(dev, > + cpumask_size(), GFP_KERNEL); > + if (!range->irqs[i].cpu_mask) > + return -ENOMEM; > + > + mask = (oirq.args[2] & 0x0000ff00) >> 8; > + for_each_set_bit(bit, &mask, BITS_PER_LONG) > + cpumask_set_cpu(bit, range->irqs[i].cpu_mask); > + } > } > > range->num_irqs = min(range->num_irqs, range->num_queues); >

7 years, 1 month

3
2
0 0

Re: [PATCH] soc: ti: QMSS: Fix usage of irq_set_affinity_hint

by Santosh Shilimkar

Hi Arnd, Olof, On 10/30/2018 4:11 AM, Marc Zyngier wrote: > The Keystone QMSS driver is pretty damaged, in the sense that it > does things like this: > > irq_set_affinity_hint(irq, to_cpumask(&cpu_map)); > > where cpu_map is a local variable. As we leave the function, this > will point to nowhere-land, and things will end-up badly. > > Instead, let's use a proper cpumask that gets allocated, giving > the driver a chance to actually work with things like irqbalance > as well as have a hypothetical 64bit future. > > Signed-off-by: Marc Zyngier <marc.zyngier(a)arm.com> > --- Could you please add this patch to your fixes branch ? FWIW, Acked-by: Santosh Shilimkar <ssantosh(a)kernel.org>

7 years, 1 month

1
0
0 0

[PATCH] usb: dwc3: core: Clean up ULPI device

by Felipe Balbi

From: Andy Shevchenko <andriy.shevchenko(a)linux.intel.com> If dwc3_core_init_mode() fails with deferred probe, next probe fails on sysfs with sysfs: cannot create duplicate filename '/devices/pci0000:00/0000:00:11.0/dwc3.0.auto/dwc3.0.auto.ulpi' To avoid this failure, clean up ULPI device. Cc: <stable(a)vger.kernel.org> Signed-off-by: Andy Shevchenko <andriy.shevchenko(a)linux.intel.com> Signed-off-by: Felipe Balbi <felipe.balbi(a)linux.intel.com> --- drivers/usb/dwc3/core.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/usb/dwc3/core.c b/drivers/usb/dwc3/core.c index 88c80fcc39f5..fec97465ccac 100644 --- a/drivers/usb/dwc3/core.c +++ b/drivers/usb/dwc3/core.c @@ -1499,6 +1499,7 @@ static int dwc3_probe(struct platform_device *pdev) err5: dwc3_event_buffers_cleanup(dwc); + dwc3_ulpi_exit(dwc); err4: dwc3_free_scratch_buffers(dwc); -- 2.19.1 --------------------------------------------------------------------- Intel Finland Oy Registered Address: PL 281, 00181 Helsinki Business Identity Code: 0357606 - 4 Domiciled in Helsinki This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). Any review or distribution by others is strictly prohibited. If you are not the intended recipient, please contact the sender and delete all copies.

7 years, 1 month

1
1
0 0

[PATCH AUTOSEL 3.18 1/5] sparc: Fix single-pcr perf event counter management.

by Sasha Levin

From: "David S. Miller" <davem(a)davemloft.net> [ Upstream commit cfdc3170d214046b9509183fe9b9544dc644d40b ] It is important to clear the hw->state value for non-stopped events when they are added into the PMU. Otherwise when the event is scheduled out, we won't read the counter because HES_UPTODATE is still set. This breaks 'perf stat' and similar use cases, causing all the events to show zero. This worked for multi-pcr because we make explicit sparc_pmu_start() calls in calculate_multiple_pcrs(). calculate_single_pcr() doesn't do this because the idea there is to accumulate all of the counter settings into the single pcr value. So we have to add explicit hw->state handling there. Like x86, we use the PERF_HES_ARCH bit to track truly stopped events so that we don't accidently start them on a reload. Related to all of this, sparc_pmu_start() is missing a userpage update so add it. Signed-off-by: David S. Miller <davem(a)davemloft.net> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- arch/sparc/kernel/perf_event.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c index af53c25da2e7..8536532970bb 100644 --- a/arch/sparc/kernel/perf_event.c +++ b/arch/sparc/kernel/perf_event.c @@ -919,6 +919,8 @@ static void read_in_all_counters(struct cpu_hw_events *cpuc) sparc_perf_event_update(cp, &cp->hw, cpuc->current_idx[i]); cpuc->current_idx[i] = PIC_NO_INDEX; + if (cp->hw.state & PERF_HES_STOPPED) + cp->hw.state |= PERF_HES_ARCH; } } } @@ -951,10 +953,12 @@ static void calculate_single_pcr(struct cpu_hw_events *cpuc) enc = perf_event_get_enc(cpuc->events[i]); cpuc->pcr[0] &= ~mask_for_index(idx); - if (hwc->state & PERF_HES_STOPPED) + if (hwc->state & PERF_HES_ARCH) { cpuc->pcr[0] |= nop_for_index(idx); - else + } else { cpuc->pcr[0] |= event_encoding(enc, idx); + hwc->state = 0; + } } out: cpuc->pcr[0] |= cpuc->event[0]->hw.config_base; @@ -980,6 +984,9 @@ static void calculate_multiple_pcrs(struct cpu_hw_events *cpuc) cpuc->current_idx[i] = idx; + if (cp->hw.state & PERF_HES_ARCH) + continue; + sparc_pmu_start(cp, PERF_EF_RELOAD); } out: @@ -1071,6 +1078,8 @@ static void sparc_pmu_start(struct perf_event *event, int flags) event->hw.state = 0; sparc_pmu_enable_event(cpuc, &event->hw, idx); + + perf_event_update_userpage(event); } static void sparc_pmu_stop(struct perf_event *event, int flags) @@ -1363,9 +1372,9 @@ static int sparc_pmu_add(struct perf_event *event, int ef_flags) cpuc->events[n0] = event->hw.event_base; cpuc->current_idx[n0] = PIC_NO_INDEX; - event->hw.state = PERF_HES_UPTODATE; + event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED; if (!(ef_flags & PERF_EF_START)) - event->hw.state |= PERF_HES_STOPPED; + event->hw.state |= PERF_HES_ARCH; /* * If group events scheduling transaction was started, -- 2.17.1

7 years, 1 month

1
4
0 0

[PATCH AUTOSEL 4.4 1/6] sparc: Fix single-pcr perf event counter management.

by Sasha Levin

From: "David S. Miller" <davem(a)davemloft.net> [ Upstream commit cfdc3170d214046b9509183fe9b9544dc644d40b ] It is important to clear the hw->state value for non-stopped events when they are added into the PMU. Otherwise when the event is scheduled out, we won't read the counter because HES_UPTODATE is still set. This breaks 'perf stat' and similar use cases, causing all the events to show zero. This worked for multi-pcr because we make explicit sparc_pmu_start() calls in calculate_multiple_pcrs(). calculate_single_pcr() doesn't do this because the idea there is to accumulate all of the counter settings into the single pcr value. So we have to add explicit hw->state handling there. Like x86, we use the PERF_HES_ARCH bit to track truly stopped events so that we don't accidently start them on a reload. Related to all of this, sparc_pmu_start() is missing a userpage update so add it. Signed-off-by: David S. Miller <davem(a)davemloft.net> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- arch/sparc/kernel/perf_event.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/arch/sparc/kernel/perf_event.c b/arch/sparc/kernel/perf_event.c index 6596f66ce112..a5d0c2f08110 100644 --- a/arch/sparc/kernel/perf_event.c +++ b/arch/sparc/kernel/perf_event.c @@ -926,6 +926,8 @@ static void read_in_all_counters(struct cpu_hw_events *cpuc) sparc_perf_event_update(cp, &cp->hw, cpuc->current_idx[i]); cpuc->current_idx[i] = PIC_NO_INDEX; + if (cp->hw.state & PERF_HES_STOPPED) + cp->hw.state |= PERF_HES_ARCH; } } } @@ -958,10 +960,12 @@ static void calculate_single_pcr(struct cpu_hw_events *cpuc) enc = perf_event_get_enc(cpuc->events[i]); cpuc->pcr[0] &= ~mask_for_index(idx); - if (hwc->state & PERF_HES_STOPPED) + if (hwc->state & PERF_HES_ARCH) { cpuc->pcr[0] |= nop_for_index(idx); - else + } else { cpuc->pcr[0] |= event_encoding(enc, idx); + hwc->state = 0; + } } out: cpuc->pcr[0] |= cpuc->event[0]->hw.config_base; @@ -987,6 +991,9 @@ static void calculate_multiple_pcrs(struct cpu_hw_events *cpuc) cpuc->current_idx[i] = idx; + if (cp->hw.state & PERF_HES_ARCH) + continue; + sparc_pmu_start(cp, PERF_EF_RELOAD); } out: @@ -1078,6 +1085,8 @@ static void sparc_pmu_start(struct perf_event *event, int flags) event->hw.state = 0; sparc_pmu_enable_event(cpuc, &event->hw, idx); + + perf_event_update_userpage(event); } static void sparc_pmu_stop(struct perf_event *event, int flags) @@ -1370,9 +1379,9 @@ static int sparc_pmu_add(struct perf_event *event, int ef_flags) cpuc->events[n0] = event->hw.event_base; cpuc->current_idx[n0] = PIC_NO_INDEX; - event->hw.state = PERF_HES_UPTODATE; + event->hw.state = PERF_HES_UPTODATE | PERF_HES_STOPPED; if (!(ef_flags & PERF_EF_START)) - event->hw.state |= PERF_HES_STOPPED; + event->hw.state |= PERF_HES_ARCH; /* * If group events scheduling transaction was started, -- 2.17.1

7 years, 1 month

1
5
0 0

[PATCH AUTOSEL 4.9 01/13] bpf: do not blindly change rlimit in reuseport net selftest

by Sasha Levin

From: Eric Dumazet <edumazet(a)google.com> [ Upstream commit 262f9d811c7608f1e74258ceecfe1fa213bdf912 ] If the current process has unlimited RLIMIT_MEMLOCK, we should should leave it as is. Fixes: 941ff6f11c02 ("bpf: fix rlimit in reuseport net selftest") Signed-off-by: John Sperbeck <jsperbeck(a)google.com> Signed-off-by: Eric Dumazet <edumazet(a)google.com> Acked-by: Daniel Borkmann <daniel(a)iogearbox.net> Signed-off-by: Daniel Borkmann <daniel(a)iogearbox.net> Signed-off-by: Sasha Levin <sashal(a)kernel.org> --- tools/testing/selftests/net/reuseport_bpf.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) diff --git a/tools/testing/selftests/net/reuseport_bpf.c b/tools/testing/selftests/net/reuseport_bpf.c index cad14cd0ea92..b5277106df1f 100644 --- a/tools/testing/selftests/net/reuseport_bpf.c +++ b/tools/testing/selftests/net/reuseport_bpf.c @@ -437,14 +437,19 @@ void enable_fastopen(void) } } -static struct rlimit rlim_old, rlim_new; +static struct rlimit rlim_old; static __attribute__((constructor)) void main_ctor(void) { getrlimit(RLIMIT_MEMLOCK, &rlim_old); - rlim_new.rlim_cur = rlim_old.rlim_cur + (1UL << 20); - rlim_new.rlim_max = rlim_old.rlim_max + (1UL << 20); - setrlimit(RLIMIT_MEMLOCK, &rlim_new); + + if (rlim_old.rlim_cur != RLIM_INFINITY) { + struct rlimit rlim_new; + + rlim_new.rlim_cur = rlim_old.rlim_cur + (1UL << 20); + rlim_new.rlim_max = rlim_old.rlim_max + (1UL << 20); + setrlimit(RLIMIT_MEMLOCK, &rlim_new); + } } static __attribute__((destructor)) void main_dtor(void) -- 2.17.1

7 years, 1 month

1
12
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror