From: ebiederm@xmission.com (Eric W. Biederman)
To: "H. Peter Anvin" <hpa@zytor.com>
Cc: Tejun Heo <tj@kernel.org>, hacklu <embedway.linux@gmail.com>,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH 2/2] x86, boot: Optimize the elf header handling.
Date: Sun, 01 Jul 2012 12:20:25 -0700 [thread overview]
Message-ID: <87sjdb8f0m.fsf@xmission.com> (raw)
In-Reply-To: <4FF098E2.3040307@zytor.com> (H. Peter Anvin's message of "Sun, 01 Jul 2012 11:37:22 -0700")
"H. Peter Anvin" <hpa@zytor.com> writes:
> On 07/01/2012 11:25 AM, Eric W. Biederman wrote:
>>
>> Well I see what is going on now.
>>
>> In the vmlinux the paddrs and the poffsests stop tracking when they
>> come to the percpu PT_LOAD segment.
>>
>> It looks like we give the percpu segment a 4K alignment in memory
>> and a 2M alignment in the file.
>>
>> It looks like something is causing ld to compute an unfortunate physical
>> load address for the percpu section.
>>
>> At first glace I would say that is a bug. Either we need the alignment
>> or we don't. But if we need the alignment we certainly need it to
>> be respected for the real memory locations of the init symbols..
>>
>
> If we don't need it, I think we can use -z max-page-size=4096, but we
> use the PMD alignment for percpu on x86-64; Tejun, does that apply to
> the .data..percpu section in the executable as well?
Grumble Grumble.
If I force a 2M alignment, poffsets and paddrs align and my previous
patches work. Although I don't expect that to be more than a testing
hack.
Meanwhile if I force a 4K max page size for most purposes things look
better from a file offset perspective. But the data segment is now less
aligned in the file than it is in memory and my patches don't work.
Sigh.
I feel like I have stumbled into pandoras box.
Eric
diff --git a/arch/x86/kernel/vmlinux.lds.S b/arch/x86/kernel/vmlinux.lds.S
index af6fb8a..7276d44 100644
--- a/arch/x86/kernel/vmlinux.lds.S
+++ b/arch/x86/kernel/vmlinux.lds.S
@@ -187,6 +187,7 @@ SECTIONS
* output PHDR, so the next output section - .init.text - should
* start another segment - init.
*/
+ . = ALIGN(0x200000);
PERCPU_VADDR(INTERNODE_CACHE_BYTES, 0, :percpu)
#endif
> -hpa
diff --git a/arch/x86/Makefile b/arch/x86/Makefile
index 1f25214..b5b31c3 100644
--- a/arch/x86/Makefile
+++ b/arch/x86/Makefile
@@ -120,7 +120,7 @@ avx_instr := $(call as-instr,vxorps %ymm0$(comma)%ymm1$(comma)%ymm2,-DCONFIG_AS_
KBUILD_AFLAGS += $(cfi) $(cfi-sigframe) $(cfi-sections) $(asinstr) $(avx_instr)
KBUILD_CFLAGS += $(cfi) $(cfi-sigframe) $(cfi-sections) $(asinstr) $(avx_instr)
-LDFLAGS := -m elf_$(UTS_MACHINE)
+LDFLAGS := -m elf_$(UTS_MACHINE) -z max-page-size=4096
# Speed up the build
KBUILD_CFLAGS += -pipe
next prev parent reply other threads:[~2012-07-01 19:20 UTC|newest]
Thread overview: 34+ messages / expand[flat|nested] mbox.gz Atom feed top
2012-05-17 5:56 why the decompressed procedure move kernel from address 0x100000(1M) to 0x1000000(16M) +x hacklu
2012-06-02 23:48 ` Eric W. Biederman
2012-06-03 3:10 ` H. Peter Anvin
2012-06-03 8:41 ` Eric W. Biederman
2012-06-03 13:01 ` H. Peter Anvin
2012-06-03 20:30 ` [PATCH 1/2] x86, boot: Don't overlap the compressed and non-compressed image Eric W. Biederman
2012-06-03 20:32 ` [PATCH 2/2] x86, boot: Optimize the elf header handling Eric W. Biederman
2012-07-01 14:56 ` [tip:x86/boot] " tip-bot for Eric W. Biederman
2012-07-01 15:04 ` [PATCH 2/2] " H. Peter Anvin
2012-07-01 15:34 ` H. Peter Anvin
2012-07-01 16:08 ` Eric W. Biederman
2012-07-01 16:11 ` H. Peter Anvin
2012-07-01 16:26 ` Eric W. Biederman
2012-07-01 16:44 ` H. Peter Anvin
2012-07-01 17:09 ` Eric W. Biederman
2012-07-01 17:15 ` H. Peter Anvin
2012-07-01 18:25 ` Eric W. Biederman
2012-07-01 18:37 ` H. Peter Anvin
2012-07-01 19:20 ` Eric W. Biederman [this message]
2012-07-01 19:23 ` H. Peter Anvin
2012-07-01 20:40 ` Eric W. Biederman
2012-07-01 20:52 ` H. Peter Anvin
2012-07-09 6:50 ` Eric W. Biederman
2012-07-09 6:52 ` [PATCH 1/4] x86 boot: Jump to the entry point address in the elf header Eric W. Biederman
2012-07-09 6:53 ` [PATCH 2/4] x86 boot: Optimize the elf header handling Eric W. Biederman
2012-07-09 6:55 ` [PATCH 3/4] x86 boot: When building vmlinux.bin properly precompute the memory image Eric W. Biederman
2012-07-09 6:56 ` [PATCH 4/4] x86 boot: Tell ld the kernel doesn't want 2MB file offset alignment Eric W. Biederman
2012-07-09 6:59 ` [PATCH 1/4] x86 boot: Jump to the entry point address in the elf header Eric W. Biederman
2012-07-02 16:56 ` [PATCH 2/2] x86, boot: Optimize the elf header handling Tejun Heo
2012-07-09 7:03 ` Eric W. Biederman
2012-07-01 2:23 ` [PATCH 1/2] x86, boot: Don't overlap the compressed and non-compressed image Eric W. Biederman
2012-07-01 2:32 ` H. Peter Anvin
2012-07-01 5:22 ` Eric W. Biederman
2012-07-01 14:55 ` [tip:x86/boot] x86, boot: Don' t " tip-bot for Eric W. Biederman
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=87sjdb8f0m.fsf@xmission.com \
--to=ebiederm@xmission.com \
--cc=embedway.linux@gmail.com \
--cc=hpa@zytor.com \
--cc=linux-kernel@vger.kernel.org \
--cc=tj@kernel.org \
/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).