From: Shile Zhang <[email protected]>
Hi,
I found the unwind_init taken long time (more than 90ms) in kernel
booting, mainly spent on sorting the two ORC unwind tables, orc_unwind
and orc_unwind_ip.
I also noticed that this issued has reported and discussed last year:
https://lkml.org/lkml/2018/10/8/342
But seems no final solution until now, I tried to sort the ORC tables at
build time, followed the helpful hints from Josh and Ingo in that thread.
And mainly referred the implementation of 'sortextable' tool:
https://lore.kernel.org/linux-mips/[email protected]/
What I did:
- Add a Kconfig to control build-time sorting or runtime sorting;
- Referred 'sortextable', create a similar helper tool 'sortorctable',
help to sort the ORC unwind tables at vmlinux link process.
One potential improvement is to sort the module ORC tables in future.
Thanks!
Shile Zhang (4):
scripts: Add sortorctable to sort ORC unwind tables
kbuild: Sort ORC unwind tables in vmlinux link process
x86/unwind/orc: Skip sorting if BUILDTIME_ORCTABLE_SORT is configured
x86/Kconfig: Add a Kconfig option to sort ORC tables at build time
arch/x86/Kconfig.debug | 9 ++
arch/x86/kernel/unwind_orc.c | 2 +
scripts/.gitignore | 1 +
scripts/Makefile | 2 +
scripts/link-vmlinux.sh | 10 ++
scripts/sortorctable.c | 246 +++++++++++++++++++++++++++++++++++
scripts/sortorctable.h | 25 ++++
7 files changed, 295 insertions(+)
create mode 100644 scripts/sortorctable.c
create mode 100644 scripts/sortorctable.h
--
2.24.0.rc2
From: Shile Zhang <[email protected]>
Sort orc_unwind and orc_unwind_ip tables at build time instead of runtime
in boot pharse can save more boot time.
Signed-off-by: Shile Zhang <[email protected]>
---
scripts/.gitignore | 1 +
scripts/Makefile | 2 +
scripts/sortorctable.c | 246 +++++++++++++++++++++++++++++++++++++++++
scripts/sortorctable.h | 25 +++++
4 files changed, 274 insertions(+)
create mode 100644 scripts/sortorctable.c
create mode 100644 scripts/sortorctable.h
diff --git a/scripts/.gitignore b/scripts/.gitignore
index 17f8cef88fa8..52976f32f974 100644
--- a/scripts/.gitignore
+++ b/scripts/.gitignore
@@ -12,3 +12,4 @@ asn1_compiler
extract-cert
sign-file
insert-sys-cert
+sortorctable
diff --git a/scripts/Makefile b/scripts/Makefile
index 3e86b300f5a1..51b2d465f042 100644
--- a/scripts/Makefile
+++ b/scripts/Makefile
@@ -16,12 +16,14 @@ hostprogs-$(CONFIG_LOGO) += pnmtologo
hostprogs-$(CONFIG_VT) += conmakehash
hostprogs-$(BUILD_C_RECORDMCOUNT) += recordmcount
hostprogs-$(CONFIG_BUILDTIME_EXTABLE_SORT) += sortextable
+hostprogs-$(CONFIG_BUILDTIME_ORCTABLE_SORT) += sortorctable
hostprogs-$(CONFIG_ASN1) += asn1_compiler
hostprogs-$(CONFIG_MODULE_SIG_FORMAT) += sign-file
hostprogs-$(CONFIG_SYSTEM_TRUSTED_KEYRING) += extract-cert
hostprogs-$(CONFIG_SYSTEM_EXTRA_CERTIFICATE) += insert-sys-cert
HOSTCFLAGS_sortextable.o = -I$(srctree)/tools/include
+HOSTCFLAGS_sortorctable.o = -I$(srctree)/tools/include
HOSTCFLAGS_asn1_compiler.o = -I$(srctree)/include
HOSTLDLIBS_sign-file = -lcrypto
HOSTLDLIBS_extract-cert = -lcrypto
diff --git a/scripts/sortorctable.c b/scripts/sortorctable.c
new file mode 100644
index 000000000000..39be47d1d296
--- /dev/null
+++ b/scripts/sortorctable.c
@@ -0,0 +1,246 @@
+// SPDX-License-Identifier: GPL-2.0
+/*
+ * sortorctable: used to sort the ORC unwind tables
+ *
+ * Strategy: alter the vmlinux file in-place
+ *
+ * Some code taken out of lib/sort.c and arch/x86/kernel/unwind_orc.c.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <string.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <sys/types.h>
+#include <sys/stat.h>
+#include <string.h>
+#include <errno.h>
+#include <sys/mman.h>
+#include <elf.h>
+
+#include "sortorctable.h"
+
+int *cur_orc_ip_table;
+struct orc_entry *cur_orc_table;
+
+/**
+ * sort - sort an array of elements
+ * @base: pointer to data to sort
+ * @num: number of elements
+ * @size: size of each element
+ * @cmp_func: pointer to comparison function
+ * @swap_func: pointer to swap function
+ *
+ * This function does a heapsort on the given array. You may provide a
+ * swap_func function optimized to your element type.
+ *
+ * Sorting time is O(n log n) both on average and worst-case. While
+ * qsort is about 20% faster on average, it suffers from exploitable
+ * O(n*n) worst-case behavior and extra memory requirements that make
+ * it less suitable for kernel use.
+ */
+static void sort(void *base, size_t num, size_t size,
+ int (*cmp_func)(const void *, const void *),
+ void (*swap_func)(void *, void *, int size))
+{
+ /* pre-scale counters for performance */
+ int i = (num/2 - 1) * size, n = num * size, c, r;
+
+ /* heapify */
+ for ( ; i >= 0; i -= size) {
+ for (r = i; r * 2 + size < n; r = c) {
+ c = r * 2 + size;
+ if (c < n - size &&
+ cmp_func(base + c, base + c + size) < 0)
+ c += size;
+ if (cmp_func(base + r, base + c) >= 0)
+ break;
+ swap_func(base + r, base + c, size);
+ }
+ }
+
+ /* sort */
+ for (i = n - size; i > 0; i -= size) {
+ swap_func(base, base + i, size);
+ for (r = 0; r * 2 + size < i; r = c) {
+ c = r * 2 + size;
+ if (c < i - size &&
+ cmp_func(base + c, base + c + size) < 0)
+ c += size;
+ if (cmp_func(base + r, base + c) >= 0)
+ break;
+ swap_func(base + r, base + c, size);
+ }
+ }
+}
+
+static inline unsigned long orc_ip(const int *ip)
+{
+ return (unsigned long)ip + *ip;
+}
+
+static void orc_sort_swap(void *_a, void *_b, int size)
+{
+ struct orc_entry *orc_a, *orc_b;
+ struct orc_entry orc_tmp;
+ int *a = _a, *b = _b, tmp;
+ int delta = _b - _a;
+
+ /* Swap the .orc_unwind_ip entries: */
+ tmp = *a;
+ *a = *b + delta;
+ *b = tmp - delta;
+
+ /* Swap the corresponding .orc_unwind entries: */
+ orc_a = cur_orc_table + (a - cur_orc_ip_table);
+ orc_b = cur_orc_table + (b - cur_orc_ip_table);
+ orc_tmp = *orc_a;
+ *orc_a = *orc_b;
+ *orc_b = orc_tmp;
+}
+
+static int orc_sort_cmp(const void *_a, const void *_b)
+{
+ struct orc_entry *orc_a;
+ const int *a = _a, *b = _b;
+ unsigned long a_val = orc_ip(a);
+ unsigned long b_val = orc_ip(b);
+
+ if (a_val > b_val)
+ return 1;
+ if (a_val < b_val)
+ return -1;
+
+ /*
+ * The "weak" section terminator entries need to always be on the left
+ * to ensure the lookup code skips them in favor of real entries.
+ * These terminator entries exist to handle any gaps created by
+ * whitelisted .o files which didn't get objtool generation.
+ */
+ orc_a = cur_orc_table + (a - cur_orc_ip_table);
+ return orc_a->sp_reg == ORC_REG_UNDEFINED && !orc_a->end ? -1 : 1;
+}
+
+/* ORC unwind only supports X86_64 */
+static int do_precheck(const char *fname, void *addr)
+{
+ Elf64_Ehdr * const ehdr = addr;
+
+ if (ehdr->e_ident[EI_DATA] != ELFDATA2LSB) {
+ fprintf(stderr, "%s: unsupported ELF data encoding %d\n",
+ fname, ehdr->e_ident[EI_DATA]);
+ return 1;
+ }
+
+ if (memcmp(ELFMAG, ehdr->e_ident, SELFMAG) != 0
+ || (ehdr->e_type != ET_EXEC && ehdr->e_type != ET_DYN)
+ || ehdr->e_ident[EI_VERSION] != EV_CURRENT) {
+ fprintf(stderr,
+ "%s: unrecognized ET_EXEC/ET_DYN file\n", fname);
+ return 1;
+ }
+
+ if (ehdr->e_machine != EM_X86_64) {
+ fprintf(stderr, "%s: unsupported e_machine %d\n",
+ fname, ehdr->e_machine);
+ return 1;
+ }
+
+ if (ehdr->e_ident[EI_CLASS] != ELFCLASS64
+ || ehdr->e_ehsize != sizeof(Elf64_Ehdr)
+ || ehdr->e_shentsize != sizeof(Elf64_Shdr)) {
+ fprintf(stderr, "%s: unrecognized ELF class %d\n",
+ fname, ehdr->e_ident[EI_CLASS]);
+ return 1;
+ }
+
+ return 0;
+}
+
+static int do_sort(const char *fname, void *addr)
+{
+ unsigned int orc_size, orc_ip_size, num_entries;
+ const Elf64_Shdr *s, *_orc = NULL, *_orc_ip = NULL;
+ Elf64_Ehdr * const ehdr = (Elf64_Ehdr *)addr;
+ Elf64_Shdr * const shdr = (Elf64_Shdr *)((void *)ehdr + ehdr->e_shoff);
+ char *secstrings = (void *)ehdr + shdr[ehdr->e_shstrndx].sh_offset;
+
+ for (s = shdr; s < shdr + ehdr->e_shnum; s++) {
+ if (!strcmp(".orc_unwind_ip", secstrings + s->sh_name))
+ _orc_ip = s;
+ if (!strcmp(".orc_unwind", secstrings + s->sh_name))
+ _orc = s;
+ }
+
+ if (!_orc_ip || !_orc) {
+ fprintf(stderr,
+ "%s: cannot find ORC unwind tables\n", fname);
+ return 1;
+ }
+
+ orc_size = _orc->sh_size;
+ orc_ip_size = _orc_ip->sh_size;
+ num_entries = orc_ip_size / sizeof(int);
+ cur_orc_table = (struct orc_entry *)((void *)ehdr + _orc->sh_offset);
+ cur_orc_ip_table = (int *)((void *)ehdr + _orc_ip->sh_offset);
+
+ if (orc_ip_size % sizeof(int) != 0
+ || orc_size % sizeof(struct orc_entry) != 0
+ || num_entries != orc_size / sizeof(struct orc_entry)) {
+ fprintf(stderr,
+ "%s: wrong ORC unwind table entries number\n", fname);
+ return 1;
+ }
+
+ sort(cur_orc_ip_table, num_entries, sizeof(int), orc_sort_cmp,
+ orc_sort_swap);
+
+ return 0;
+}
+
+int main(int argc, char *argv[])
+{
+ int ret = 0;
+ int fd;
+ char *fname = NULL;
+ struct stat sb;
+ void *addr = NULL;
+
+ if (argc != 2) {
+ fprintf(stderr, "usage: sortorctable vmlinux\n");
+ return 0;
+ }
+
+ fname = argv[1];
+ fd = open(fname, O_RDWR);
+ if (fd < 0 || fstat(fd, &sb) < 0) {
+ perror(fname);
+ return errno;
+ }
+
+ if (!S_ISREG(sb.st_mode)) {
+ fprintf(stderr, "'%s': not a regular file\n", fname);
+ close(fd);
+ return 1;
+ }
+
+ addr = mmap(0, sb.st_size, PROT_READ|PROT_WRITE, MAP_SHARED, fd, 0);
+ if (addr == MAP_FAILED) {
+ fprintf(stderr, "'%s': mmap failed\n", fname);
+ close(fd);
+ return 1;
+ }
+
+ ret = do_precheck(fname, addr);
+ if (ret)
+ goto out;
+
+ ret = do_sort(fname, addr);
+
+out:
+ munmap(addr, sb.st_size);
+ close(fd);
+
+ return ret;
+}
diff --git a/scripts/sortorctable.h b/scripts/sortorctable.h
new file mode 100644
index 000000000000..8a7d3bd8b01b
--- /dev/null
+++ b/scripts/sortorctable.h
@@ -0,0 +1,25 @@
+/* SPDX-License-Identifier: GPL-2.0 */
+/*
+ * Copyright (C) 2019 Shile Zhang <[email protected]>
+ *
+ * This code was taken out of arch/x86/include/asm/orc_types.h written by:
+ * Copyright (C) 2017 Josh Poimboeuf <[email protected]>
+ */
+
+#ifndef _SORTORCTABLE_H_
+#define _SORTORCTABLE_H_
+
+#include <linux/types.h>
+
+#define ORC_REG_UNDEFINED 0
+
+struct orc_entry {
+ s16 sp_offset;
+ s16 bp_offset;
+ unsigned sp_reg:4;
+ unsigned bp_reg:4;
+ unsigned type:2;
+ unsigned end:1;
+} __attribute__((packed));
+
+#endif//_SORTORCTABLE_H_
--
2.24.0.rc2
On Thu, Nov 07, 2019 at 10:32:01PM +0800, [email protected] wrote:
> From: Shile Zhang <[email protected]>
>
> Hi,
>
> I found the unwind_init taken long time (more than 90ms) in kernel
> booting, mainly spent on sorting the two ORC unwind tables, orc_unwind
> and orc_unwind_ip.
>
> I also noticed that this issued has reported and discussed last year:
> https://lkml.org/lkml/2018/10/8/342
> But seems no final solution until now, I tried to sort the ORC tables at
> build time, followed the helpful hints from Josh and Ingo in that thread.
> And mainly referred the implementation of 'sortextable' tool:
> https://lore.kernel.org/linux-mips/[email protected]/
>
> What I did:
>
> - Add a Kconfig to control build-time sorting or runtime sorting;
> - Referred 'sortextable', create a similar helper tool 'sortorctable',
> help to sort the ORC unwind tables at vmlinux link process.
What is the build-time cost for doing this? The link phase is already a
fairly big bottleneck for building a kernel.
Can sort{ex,orc}table() be ran concurrently? Do they want to be the same
(threaded) tool?
On Thu, Nov 07, 2019 at 10:32:01PM +0800, [email protected] wrote:
> From: Shile Zhang <[email protected]>
>
> Hi,
>
> I found the unwind_init taken long time (more than 90ms) in kernel
> booting, mainly spent on sorting the two ORC unwind tables, orc_unwind
> and orc_unwind_ip.
>
> I also noticed that this issued has reported and discussed last year:
> https://lkml.org/lkml/2018/10/8/342
> But seems no final solution until now, I tried to sort the ORC tables at
> build time, followed the helpful hints from Josh and Ingo in that thread.
> And mainly referred the implementation of 'sortextable' tool:
> https://lore.kernel.org/linux-mips/[email protected]/
>
> What I did:
>
> - Add a Kconfig to control build-time sorting or runtime sorting;
> - Referred 'sortextable', create a similar helper tool 'sortorctable',
> help to sort the ORC unwind tables at vmlinux link process.
>
> One potential improvement is to sort the module ORC tables in future.
>
> Thanks!
Thanks a lot for working on this!
I'd say the new config option isn't needed. The runtime ORC sorting
logic is unconditionally bad and the code should just be removed. I saw
recently that it's one of the main offenders for boot time latency.
I also agree with Peter that we should try to reduce the link-time
penalty as much as possible. But it's a necessary evil to a certain
extent.
--
Josh
On 2019/11/7 23:22, Peter Zijlstra wrote:
> On Thu, Nov 07, 2019 at 10:32:01PM +0800, [email protected] wrote:
>> From: Shile Zhang <[email protected]>
>>
>> Hi,
>>
>> I found the unwind_init taken long time (more than 90ms) in kernel
>> booting, mainly spent on sorting the two ORC unwind tables, orc_unwind
>> and orc_unwind_ip.
>>
>> I also noticed that this issued has reported and discussed last year:
>> https://lkml.org/lkml/2018/10/8/342
>> But seems no final solution until now, I tried to sort the ORC tables at
>> build time, followed the helpful hints from Josh and Ingo in that thread.
>> And mainly referred the implementation of 'sortextable' tool:
>> https://lore.kernel.org/linux-mips/[email protected]/
>>
>> What I did:
>>
>> - Add a Kconfig to control build-time sorting or runtime sorting;
>> - Referred 'sortextable', create a similar helper tool 'sortorctable',
>> help to sort the ORC unwind tables at vmlinux link process.
> What is the build-time cost for doing this? The link phase is already a
> fairly big bottleneck for building a kernel.
Hi, Peter, Thanks for your kindly reply!
On my test env, the build-time sort spend about 100ms, which is similar
to runtime sorting due to the same sorting code. Of course there are few
compiling cost in sortorctable tool itself. I think the overall cost of
this build-time sorting is not so much.
I agree with you that the link time of vmlinux shoud be optimized.
But IMHO, for one kernel product release, the kernel building is once
for all. So put the sorting in build time can save boot time for
customer, for each booting. I think this is significant for boot time
sensitive products, such as embedded devices in IoT, or VM in Cloud.
> Can sort{ex,orc}table() be ran concurrently? Do they want to be the same
> (threaded) tool?
I think it is possible to do those sort work concurrently, likes
deferred memory init which is big boot time speed up.
But I don't know if the exception table and ORC unwind tables can be
deferred, due to those tables might be used in early boot time, for
early exception handling and early debugging. I'm not familiar with that.
IMO, the init works in kernel boot time should do the necessary
runtime-depends initialization, which cannot done out-of-bands. For
exception, ORC unwind like tables, which does not depends on the runtime
ENV. It can/should be ready in building time. IOW, this kind of "setup
cost" should not put on customer.
Thanks again!
On 2019/11/7 23:46, Josh Poimboeuf wrote:
> On Thu, Nov 07, 2019 at 10:32:01PM +0800, [email protected] wrote:
>> From: Shile Zhang <[email protected]>
>>
>> Hi,
>>
>> I found the unwind_init taken long time (more than 90ms) in kernel
>> booting, mainly spent on sorting the two ORC unwind tables, orc_unwind
>> and orc_unwind_ip.
>>
>> I also noticed that this issued has reported and discussed last year:
>> https://lkml.org/lkml/2018/10/8/342
>> But seems no final solution until now, I tried to sort the ORC tables at
>> build time, followed the helpful hints from Josh and Ingo in that thread.
>> And mainly referred the implementation of 'sortextable' tool:
>> https://lore.kernel.org/linux-mips/[email protected]/
>>
>> What I did:
>>
>> - Add a Kconfig to control build-time sorting or runtime sorting;
>> - Referred 'sortextable', create a similar helper tool 'sortorctable',
>> help to sort the ORC unwind tables at vmlinux link process.
>>
>> One potential improvement is to sort the module ORC tables in future.
>>
>> Thanks!
> Thanks a lot for working on this!
>
> I'd say the new config option isn't needed. The runtime ORC sorting
> logic is unconditionally bad and the code should just be removed. I saw
> recently that it's one of the main offenders for boot time latency.
Hi, Josh, Thanks very much for your quickly response!
I'll refactor the code as your advice.
Yes, the run-time sorting cost is bigger for currently Cloud products,
such as serverless products, which needs boot up ASAP.
>
> I also agree with Peter that we should try to reduce the link-time
> penalty as much as possible. But it's a necessary evil to a certain
> extent.
>
agree!
Thanks again and looking forwards more advice from you!
On Fri, Nov 08, 2019 at 09:42:55AM +0800, Shile Zhang wrote:
> > Can sort{ex,orc}table() be ran concurrently? Do they want to be the same
> > (threaded) tool?
> I think it is possible to do those sort work concurrently, likes deferred
> memory init which is big boot time speed up.
> But I don't know if the exception table and ORC unwind tables can be
> deferred, due to those tables might be used in early boot time, for early
> exception handling and early debugging. I'm not familiar with that.
I meant at link time, run both sorts concurrently such that we only have
to wait for the longest, instead of the sum of them.
They're not changing the same part of the ELF file, so it should be
possible to have one tool have multiple threads, each sorting a
different table.
Aside from the .ex_table and ORC there's also .jump_table that wants
sorting (see jump_label_sort_entries()).
I agree that doing it at link time makes sense, I just hate to do all
this sorting in sequence and blowing up the link time. I don't build for
customers, I build for single use boot and linking _SUCKS_.
On Fri, Nov 08, 2019 at 10:21:36AM +0100, Peter Zijlstra wrote:
> On Fri, Nov 08, 2019 at 09:42:55AM +0800, Shile Zhang wrote:
>
> > > Can sort{ex,orc}table() be ran concurrently? Do they want to be the same
> > > (threaded) tool?
>
> > I think it is possible to do those sort work concurrently, likes deferred
> > memory init which is big boot time speed up.
> > But I don't know if the exception table and ORC unwind tables can be
> > deferred, due to those tables might be used in early boot time, for early
> > exception handling and early debugging. I'm not familiar with that.
>
> I meant at link time, run both sorts concurrently such that we only have
> to wait for the longest, instead of the sum of them.
>
> They're not changing the same part of the ELF file, so it should be
> possible to have one tool have multiple threads, each sorting a
> different table.
>
> Aside from the .ex_table and ORC there's also .jump_table that wants
> sorting (see jump_label_sort_entries()).
Oh, and I'll be adding .static_call_sites soon, see:
https://lkml.kernel.org/r/[email protected]
(I should repost that)
That gives us 4 tables to sort which we can do concurrently in 4
threads.
> I agree that doing it at link time makes sense, I just hate to do all
> this sorting in sequence and blowing up the link time. I don't build for
> customers, I build for single use boot and linking _SUCKS_.
On 2019/11/8 17:25, Peter Zijlstra wrote:
> On Fri, Nov 08, 2019 at 10:21:36AM +0100, Peter Zijlstra wrote:
>> On Fri, Nov 08, 2019 at 09:42:55AM +0800, Shile Zhang wrote:
>>
>>>> Can sort{ex,orc}table() be ran concurrently? Do they want to be the same
>>>> (threaded) tool?
>>> I think it is possible to do those sort work concurrently, likes deferred
>>> memory init which is big boot time speed up.
>>> But I don't know if the exception table and ORC unwind tables can be
>>> deferred, due to those tables might be used in early boot time, for early
>>> exception handling and early debugging. I'm not familiar with that.
>> I meant at link time, run both sorts concurrently such that we only have
>> to wait for the longest, instead of the sum of them.
>>
>> They're not changing the same part of the ELF file, so it should be
>> possible to have one tool have multiple threads, each sorting a
>> different table.
>>
>> Aside from the .ex_table and ORC there's also .jump_table that wants
>> sorting (see jump_label_sort_entries()).
> Oh, and I'll be adding .static_call_sites soon, see:
>
> https://lkml.kernel.org/r/[email protected]
>
> (I should repost that)
>
> That gives us 4 tables to sort which we can do concurrently in 4
> threads.
I got your point now.
I'll try to rework the sort tool to sort all tables concurrently in one
tool with multiple-threads.
Thanks for your advice!
>> I agree that doing it at link time makes sense, I just hate to do all
>> this sorting in sequence and blowing up the link time. I don't build for
>> customers, I build for single use boot and linking _SUCKS_.