linux-riscv.lists.infradead.org archive mirror
 help / color / mirror / Atom feed
From: Amma Lee <lixiaoyun@binary-semi.com>
To: paul.walmsley@sifive.com, palmer@dabbelt.com, aou@eecs.berkeley.edu
Cc: lixiaoyun@binary-semi.com, xiezx@binary-semi.com,
	linux-riscv@lists.infradead.org, linux-kernel@vger.kernel.org
Subject: [PATCH] riscv: optimize ELF relocation function in riscv
Date: Mon,  3 Jul 2023 11:32:12 +0800	[thread overview]
Message-ID: <1688355132-62933-1-git-send-email-lixiaoyun@binary-semi.com> (raw)

The patch can optimize the running times of insmod command by modify ELF
relocation function.
In the 5.10 and latest kernel, when install the riscv ELF drivers which
contains multiple symbol table items to be relocated, kernel takes a lot
of time to execute the relocation. For example, we install a 3+MB driver
need 180+s.
We focus on the riscv architecture handle R_RISCV_HI20 and R_RISCV_LO20
type items relocation function in the arch\riscv\kernel\module.c and
find that there are two-loops in the function. If we modify the begin
number in the second for-loops iteration, we could save significant time
for installation. We install the same 3+MB driver could just need 2s.

Signed-off-by: Amma Lee <lixiaoyun@binary-semi.com>
---
 arch/riscv/kernel/module.c | 67 +++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 64 insertions(+), 3 deletions(-)

diff --git a/arch/riscv/kernel/module.c b/arch/riscv/kernel/module.c
index 7c651d5..b8df144 100755
--- a/arch/riscv/kernel/module.c
+++ b/arch/riscv/kernel/module.c
@@ -345,13 +345,13 @@ int apply_relocate_add(Elf_Shdr *sechdrs, const char *strtab,
 	int (*handler)(struct module *me, u32 *location, Elf_Addr v);
 	Elf_Sym *sym;
 	u32 *location;
-	unsigned int i, type;
+	unsigned int i, type, j_idx;
 	Elf_Addr v;
 	int res;
 
 	pr_debug("Applying relocate section %u to %u\n", relsec,
 	       sechdrs[relsec].sh_info);
-
+	j_idx = 0;
 	for (i = 0; i < sechdrs[relsec].sh_size / sizeof(*rel); i++) {
 		/* This is where to make the change */
 		location = (void *)sechdrs[sechdrs[relsec].sh_info].sh_addr
@@ -386,7 +386,15 @@ int apply_relocate_add(Elf_Shdr *sechdrs, const char *strtab,
 		if (type == R_RISCV_PCREL_LO12_I || type == R_RISCV_PCREL_LO12_S) {
 			unsigned int j;
 
-			for (j = 0; j < sechdrs[relsec].sh_size / sizeof(*rel); j++) {
+			/*In the second for-loops, each traversal for j is
+			 * starts from 0 to the symbol table item index which
+			 * is detected. By the tool "readelf", we find that all
+			 * the symbol table items about R_RISCV_PCREL_HI20 type
+			 * are incrementally added in order. It means that we
+			 * could interate the j with the previous loop end
+			 * value(j_idx) as the begin number in the next loop;
+			 */
+			for (j = j_idx; j < sechdrs[relsec].sh_size / sizeof(*rel); j++) {
 				unsigned long hi20_loc =
 					sechdrs[sechdrs[relsec].sh_info].sh_addr
 					+ rel[j].r_offset;
@@ -420,11 +428,64 @@ int apply_relocate_add(Elf_Shdr *sechdrs, const char *strtab,
 				}
 			}
 			if (j == sechdrs[relsec].sh_size / sizeof(*rel)) {
+				if (j_idx == 0) {
 				pr_err(
 				  "%s: Can not find HI20 relocation information\n",
 				  me->name);
 				return -EINVAL;
+}
+
+				/*If the last j-loop have been traversed to the
+				 * maximum value but never match the
+				 * corresponding symbol relocation item, the
+				 * j-loop will execute the second loop which
+				 * is begin from 0 to the prerious index (j_idx)
+				 * unless the previous j_idx == 0;
+				 * */
+				for (j = 0; j < j_idx; j++) {
+				unsigned long hi20_loc =
+				sechdrs[sechdrs[relsec].sh_info].sh_addr
+						+ rel[j].r_offset;
+				u32 hi20_type = ELF_RISCV_R_TYPE(rel[j].r_info);
+
+				/* Find the corresponding HI20 relocation entry */
+				if (hi20_loc == sym->st_value
+					&& (hi20_type == R_RISCV_PCREL_HI20
+					|| hi20_type == R_RISCV_GOT_HI20)) {
+					s32 hi20, lo12;
+					Elf_Sym *hi20_sym =
+						(Elf_Sym *)sechdrs[symindex].sh_addr
+						+ ELF_RISCV_R_SYM(rel[j].r_info);
+					unsigned long hi20_sym_val =
+						hi20_sym->st_value
+						+ rel[j].r_addend;
+
+					/* Calculate lo12 */
+					size_t offset = hi20_sym_val - hi20_loc;
+					if (IS_ENABLED(CONFIG_MODULE_SECTIONS)
+					    && hi20_type == R_RISCV_GOT_HI20) {
+						offset = module_emit_got_entry(
+							 me, hi20_sym_val);
+						offset = offset - hi20_loc;
+					}
+					hi20 = (offset + 0x800) & 0xfffff000;
+					lo12 = offset - hi20;
+					v = lo12;
+
+					break;
+				}
 			}
+
+				if (j == j_idx) {
+					pr_err(
+						"%s: Can not find HI20 relocation information\n",
+						me->name);
+					return -EINVAL;
+				}
+			}
+
+			/* Record the previous j-loop end index */
+			j_idx = j;
 		}
 
 		res = handler(me, location, v);
-- 
1.9.1


_______________________________________________
linux-riscv mailing list
linux-riscv@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/linux-riscv

             reply	other threads:[~2023-07-03  3:33 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-07-03  3:32 Amma Lee [this message]
2023-07-03 10:35 ` [PATCH] riscv: optimize ELF relocation function in riscv Andrew Jones
2023-07-03 10:47   ` Conor Dooley
     [not found]     ` <7abf1c9c-c1b8-4b07-9be9-6a920fd5da6f.lixiaoyun@binary-semi.com>
2023-07-25  9:32       ` 回复:[PATCH] " Conor.Dooley
  -- strict thread matches above, loose matches on Subject: below --
2023-05-10 11:10 [PATCH] " Amma.Lee
2023-05-11 22:45 ` Conor Dooley

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1688355132-62933-1-git-send-email-lixiaoyun@binary-semi.com \
    --to=lixiaoyun@binary-semi.com \
    --cc=aou@eecs.berkeley.edu \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-riscv@lists.infradead.org \
    --cc=palmer@dabbelt.com \
    --cc=paul.walmsley@sifive.com \
    --cc=xiezx@binary-semi.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).