lkml.org 
[lkml]   [2020]   [Jun]   [5]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
Patch in this message
/
Date
Subject[PATCH v1] arm64/module: Optimize module load time by optimizing PLT counting
From
When loading a module, module_frob_arch_sections() tries to figure out
the number of PLTs that'll be needed to handle all the RELAs. While
doing this, it tries to dedupe PLT allocations for multiple
R_AARCH64_CALL26 relocations to the same symbol. It does the same for
R_AARCH64_JUMP26 relocations too.

To make checks for duplicates easier/faster, it sorts the relocation
list by type, symbol and addend. That way, to check for a duplicate
relocation, it just needs to compare with the previous entry.

However, sorting the entire relocation array is unnecessary and
expensive (O(n log n)) because there are a lot of other relocation types
that don't need deduping or can't be deduped.

So this commit partitions the array into entries that need deduping and
those that don't. And then sorts just the part that needs deduping. And
when CONFIG_RANDOMIZE_BASE is disabled, the sorting is skipped entirely
because PLTs are not allocated for R_AARCH64_CALL26 and R_AARCH64_JUMP26
if it's disabled.

This gives significant reduction in module load time for modules with
large number of relocations with no measurable impact on modules with a
small number of relocations. In my test setup with CONFIG_RANDOMIZE_BASE
enabled, the load time for one module went down from 268ms to 100ms.
Another module went down from 143ms to 83ms.

This commit also disables the sorting if CONFIG_RANDOMIZE_BASE is
disabled because it looks like PLTs are not allocated for
R_AARCH64_CALL26 and R_AARCH64_JUMP26 if it's disabled.

Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: Saravana Kannan <saravanak@google.com>
---
arch/arm64/kernel/module-plts.c | 37 ++++++++++++++++++++++++++++++++-
1 file changed, 36 insertions(+), 1 deletion(-)

diff --git a/arch/arm64/kernel/module-plts.c b/arch/arm64/kernel/module-plts.c
index 65b08a74aec6..bf5118b3b828 100644
--- a/arch/arm64/kernel/module-plts.c
+++ b/arch/arm64/kernel/module-plts.c
@@ -253,6 +253,36 @@ static unsigned int count_plts(Elf64_Sym *syms, Elf64_Rela *rela, int num,
return ret;
}

+static bool rela_needs_dedup(Elf64_Rela *rela)
+{
+ return ELF64_R_TYPE(rela->r_info) == R_AARCH64_JUMP26
+ || ELF64_R_TYPE(rela->r_info) == R_AARCH64_CALL26;
+}
+
+/* Group the CALL26/JUMP26 relas toward the beginning of the array. */
+static int partition_dedup_relas(Elf64_Rela *rela, int numrels)
+{
+ int i = 0, j = numrels - 1;
+ Elf64_Rela t;
+
+ while (i < j) {
+ while (rela_needs_dedup(rela + i) && i < j)
+ i++;
+ while (!rela_needs_dedup(rela + j) && i < j)
+ j--;
+ if (i < j) {
+ t = *(rela + j);
+ *(rela + j) = *(rela + i);
+ *(rela + i) = t;
+ }
+ }
+ /* If the entire array needs dedup, make sure i == numrels */
+ if (rela_needs_dedup(rela + i))
+ i++;
+
+ return i;
+}
+
int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs,
char *secstrings, struct module *mod)
{
@@ -291,6 +321,7 @@ int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs,
for (i = 0; i < ehdr->e_shnum; i++) {
Elf64_Rela *rels = (void *)ehdr + sechdrs[i].sh_offset;
int numrels = sechdrs[i].sh_size / sizeof(Elf64_Rela);
+ int num_dedup_rels = 0;
Elf64_Shdr *dstsec = sechdrs + sechdrs[i].sh_info;

if (sechdrs[i].sh_type != SHT_RELA)
@@ -300,8 +331,12 @@ int module_frob_arch_sections(Elf_Ehdr *ehdr, Elf_Shdr *sechdrs,
if (!(dstsec->sh_flags & SHF_EXECINSTR))
continue;

+ if (IS_ENABLED(CONFIG_RANDOMIZE_BASE))
+ num_dedup_rels = partition_dedup_relas(rels, numrels);
/* sort by type, symbol index and addend */
- sort(rels, numrels, sizeof(Elf64_Rela), cmp_rela, NULL);
+ if (num_dedup_rels)
+ sort(rels, num_dedup_rels, sizeof(Elf64_Rela),
+ cmp_rela, NULL);

if (!str_has_prefix(secstrings + dstsec->sh_name, ".init"))
core_plts += count_plts(syms, rels, numrels,
--
2.27.0.278.ge193c7cf3a9-goog
\
 
 \ /
  Last update: 2020-06-06 00:23    [W:0.073 / U:0.140 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site