[PATCH] x86: fix 2 section mismatch warnings - find_and_reserve_crashkernel

[PATCH] x86: fix 2 section mismatch warnings - find_and_reserve_crashkernel

[Marcin Slusarz]

WARNING: vmlinux.o(.text+0xcd1f): Section mismatch in reference from the function find_and_reserve_crashkernel() to the function .init.text:find_e820_area()
The function find_and_reserve_crashkernel() references
the function __init find_e820_area().
This is often because find_and_reserve_crashkernel lacks a __init
annotation or the annotation of find_e820_area is wrong.

WARNING: vmlinux.o(.text+0xcd38): Section mismatch in reference from the function find_and_reserve_crashkernel() to the function .init.text:reserve_bootmem_generic()
The function find_and_reserve_crashkernel() references
the function __init reserve_bootmem_generic().
This is often because find_and_reserve_crashkernel lacks a __init
annotation or the annotation of reserve_bootmem_generic is wrong.

find_and_reserve_crashkernel is called from __init function (reserve_crashkernel)
and calls 2 __init functions (find_e820_area, reserve_bootmem_generic),
so mark it __init

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
---
 arch/x86/kernel/setup.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/arch/x86/kernel/setup.c b/arch/x86/kernel/setup.c
index 68b48e3..a4656ad 100644
--- a/arch/x86/kernel/setup.c
+++ b/arch/x86/kernel/setup.c
@@ -445,7 +445,7 @@ static void __init reserve_early_setup_data(void)
  * @size: Size of the crashkernel memory to reserve.
  * Returns the base address on success, and -1ULL on failure.
  */
-unsigned long long find_and_reserve_crashkernel(unsigned long long size)
+unsigned long long __init find_and_reserve_crashkernel(unsigned long long size)
 {
 	const unsigned long long alignment = 16<<20; 	/* 16M */
 	unsigned long long start = 0LL;
-- 
1.5.4.5

Re: [PATCH] x86: fix 2 section mismatch warnings - find_and_reserve_crashkernel

[Ingo Molnar]

applied these fixes of yours to tip/x86/urgent:

# f0b0d88: x86: fix 2 section mismatch warnings - find_and_reserve_crashkernel
# 844342a: x86: fix section mismatch warning - spp_getpage()
# c9d08f0: x86: fix 2 section mismatch warnings - map_high()

thanks Marcin.

	Ingo

Re: [PATCH] x86: fix 2 section mismatch warnings - find_and_reserve_crashkernel

[Ingo Molnar]

* Ingo Molnar <mingo@elte.hu> wrote:

> # 844342a: x86: fix section mismatch warning - spp_getpage()

i've reverted this one as it's wrong and can cause crashes. For example 
native_set_fixmap() gets called on 32-bit kernels from the ELF loader, 
via arch_setup_additional_pages().

	Ingo

Re: [PATCH] x86: fix 2 section mismatch warnings - find_and_reserve_crashkernel

[Marcin Slusarz]

On Wed, Aug 13, 2008 at 04:05:55PM +0200, Ingo Molnar wrote:
> 
> * Ingo Molnar <mingo@elte.hu> wrote:
> 
> > # 844342a: x86: fix section mismatch warning - spp_getpage()
> 
> i've reverted this one as it's wrong and can cause crashes. For example 
> native_set_fixmap() gets called on 32-bit kernels from the ELF loader, 
> via arch_setup_additional_pages().


$ git grep  arch_setup_additional_pages | cat
arch/powerpc/include/asm/elf.h:/* vDSO has arch_setup_additional_pages */
arch/powerpc/include/asm/elf.h:extern int arch_setup_additional_pages(struct linux_binprm *bprm,
arch/powerpc/kernel/vdso.c:int arch_setup_additional_pages(struct linux_binprm *bprm,
arch/sh/include/asm/elf.h:/* vDSO has arch_setup_additional_pages */
arch/sh/include/asm/elf.h:extern int arch_setup_additional_pages(struct linux_binprm *bprm,
arch/sh/kernel/vsyscall/vsyscall.c:int arch_setup_additional_pages(struct linux_binprm *bprm,
arch/x86/vdso/vdso32-setup.c:#define arch_setup_additional_pages        syscall32_setup_pages
arch/x86/vdso/vdso32-setup.c:int arch_setup_additional_pages(struct linux_binprm *bprm, int exstack)
arch/x86/vdso/vma.c:int arch_setup_additional_pages(struct linux_binprm *bprm, int exstack)
fs/binfmt_elf.c:        retval = arch_setup_additional_pages(bprm, executable_stack);
fs/compat_binfmt_elf.c:#ifdef   compat_arch_setup_additional_pages
fs/compat_binfmt_elf.c:#undef   arch_setup_additional_pages
fs/compat_binfmt_elf.c:#define  arch_setup_additional_pages compat_arch_setup_additional_pages
include/asm-x86/elf.h:extern int arch_setup_additional_pages(struct linux_binprm *bprm,
include/asm-x86/elf.h:#define compat_arch_setup_additional_pages        syscall32_setup_pages

$ git grep native_set_fixmap | cat
arch/x86/kernel/paravirt.c:     .set_fixmap = native_set_fixmap,
arch/x86/mm/pgtable.c:void __native_set_fixmap(enum fixed_addresses idx, pte_t pte)
arch/x86/mm/pgtable.c:void native_set_fixmap(enum fixed_addresses idx, unsigned long phys, pgprot_t flags)
arch/x86/mm/pgtable.c:  __native_set_fixmap(idx, pfn_pte(phys >> PAGE_SHIFT, flags));
arch/x86/xen/enlighten.c:       __native_set_fixmap(idx, pte);
include/asm-x86/fixmap.h:void __native_set_fixmap(enum fixed_addresses idx, pte_t pte);
include/asm-x86/fixmap.h:void native_set_fixmap(enum fixed_addresses idx,
include/asm-x86/fixmap.h:       native_set_fixmap(idx, phys, flags);

I looked at both implementations of arch_setup_additional_pages and didn't see any call to native_set_fixmap
or function which calls directly native_set_fixmap... Can you tell me what's the callchain?

Marcin

Re: [PATCH] x86: fix 2 section mismatch warnings - find_and_reserve_crashkernel

[Ingo Molnar]

* Marcin Slusarz <marcin.slusarz@gmail.com> wrote:

> I looked at both implementations of arch_setup_additional_pages and 
> didn't see any call to native_set_fixmap or function which calls 
> directly native_set_fixmap... Can you tell me what's the callchain?

the callchain would be:

[   20.116039] PM: Adding info for No Bus:vcsa1
[   20.180039] BUG: unable to handle kernel paging request at c0833e44
[   20.180039] IP: [<c0833e44>] native_set_fixmap+0x0/0x1d
[   20.180039] *pde = 36f2a163 *pte = 00833162
[   20.180039] Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
[   20.180039] Modules linked in:
[   20.180039]
[   20.180039] Pid: 1, comm: init Not tainted (2.6.27-rc3-00005-g844342a-dirty #24071)
[   20.180039] EIP: 0060:[<c0833e44>] EFLAGS: 00010216 CPU: 0
[   20.180039] EIP is at native_set_fixmap+0x0/0x1d
[   20.180039] EAX: 00000001 EBX: 00000001 ECX: 00000025 EDX: 3782e000
[   20.180039] ESI: 00000000 EDI: f671c000 EBP: f7836e68 ESP: f7836e4c
[   20.180039]  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
[   20.180039] Process init (pid: 1, ti=f7836000 task=f7870000 task.ti=f7836000)
[   20.180039] Stack: c011d580 f671c034 c018bbea 00000025 00000000 00000988 f6ef2000 f7836f1c
[   20.180039]        c01b43d6 00000003 00000008 f7836f7c f782fd10 00000003 f78b6100 08048000
[   20.180039]        00000000 00000001 f67a2500 08050b99 08050b99 00000000 00000000 00000000
[   20.180039] Call Trace:
[   20.180039]  [<c011d580>] ? arch_setup_additional_pages+0x8c/0x124
[   20.180039]  [<c018bbea>] ? try_module_get+0x8/0x2f
[   20.180039]  [<c01b43d6>] ? load_elf_binary+0xae6/0x1108
[   20.180039]  [<c0144c6d>] ? lock_release_holdtime+0x43/0x48
[   20.180039]  [<c018c0c0>] ? search_binary_handler+0x92/0x1e5
[   20.180039]  [<c01b38f0>] ? load_elf_binary+0x0/0x1108
[   20.180039]  [<c018d2d7>] ? do_execve+0x14c/0x1d8
[   20.180039]  [<c0101dd8>] ? sys_execve+0x2e/0x56
[   20.180039]  [<c01030ca>] ? syscall_call+0x7/0xb
[   20.180039]  [<c01067c8>] ? kernel_execve+0x1c/0x21
[   20.180039]  [<c0101286>] ? run_init_process+0x1c/0x1e
[   20.180039]  [<c0101333>] ? init_post+0xab/0xd4
[   20.180039]  [<c0103d07>] ? kernel_thread_helper+0x7/0x10
[   20.180039]  =======================
[   20.180039] Code:  Bad EIP value.

	Ingo

[PATCH v2] x86: fix section mismatch warning - spp_getpage()

[Marcin Slusarz]

On Fri, Aug 15, 2008 at 03:50:27PM +0200, Ingo Molnar wrote:
> 
> * Marcin Slusarz <marcin.slusarz@gmail.com> wrote:
> 
> > I looked at both implementations of arch_setup_additional_pages and 
> > didn't see any call to native_set_fixmap or function which calls 
> > directly native_set_fixmap... Can you tell me what's the callchain?
> 
> the callchain would be:
> 
> [   20.116039] PM: Adding info for No Bus:vcsa1
> [   20.180039] BUG: unable to handle kernel paging request at c0833e44
> [   20.180039] IP: [<c0833e44>] native_set_fixmap+0x0/0x1d
> [   20.180039] *pde = 36f2a163 *pte = 00833162
> [   20.180039] Oops: 0000 [#1] SMP DEBUG_PAGEALLOC
> [   20.180039] Modules linked in:
> [   20.180039]
> [   20.180039] Pid: 1, comm: init Not tainted (2.6.27-rc3-00005-g844342a-dirty #24071)
> [   20.180039] EIP: 0060:[<c0833e44>] EFLAGS: 00010216 CPU: 0
> [   20.180039] EIP is at native_set_fixmap+0x0/0x1d
> [   20.180039] EAX: 00000001 EBX: 00000001 ECX: 00000025 EDX: 3782e000
> [   20.180039] ESI: 00000000 EDI: f671c000 EBP: f7836e68 ESP: f7836e4c
> [   20.180039]  DS: 007b ES: 007b FS: 00d8 GS: 0000 SS: 0068
> [   20.180039] Process init (pid: 1, ti=f7836000 task=f7870000 task.ti=f7836000)
> [   20.180039] Stack: c011d580 f671c034 c018bbea 00000025 00000000 00000988 f6ef2000 f7836f1c
> [   20.180039]        c01b43d6 00000003 00000008 f7836f7c f782fd10 00000003 f78b6100 08048000
> [   20.180039]        00000000 00000001 f67a2500 08050b99 08050b99 00000000 00000000 00000000
> [   20.180039] Call Trace:
> [   20.180039]  [<c011d580>] ? arch_setup_additional_pages+0x8c/0x124
> [   20.180039]  [<c018bbea>] ? try_module_get+0x8/0x2f
> [   20.180039]  [<c01b43d6>] ? load_elf_binary+0xae6/0x1108
> [   20.180039]  [<c0144c6d>] ? lock_release_holdtime+0x43/0x48
> [   20.180039]  [<c018c0c0>] ? search_binary_handler+0x92/0x1e5
> [   20.180039]  [<c01b38f0>] ? load_elf_binary+0x0/0x1108
> [   20.180039]  [<c018d2d7>] ? do_execve+0x14c/0x1d8
> [   20.180039]  [<c0101dd8>] ? sys_execve+0x2e/0x56
> [   20.180039]  [<c01030ca>] ? syscall_call+0x7/0xb
> [   20.180039]  [<c01067c8>] ? kernel_execve+0x1c/0x21
> [   20.180039]  [<c0101286>] ? run_init_process+0x1c/0x1e
> [   20.180039]  [<c0101333>] ? init_post+0xab/0xd4
> [   20.180039]  [<c0103d07>] ? kernel_thread_helper+0x7/0x10
> [   20.180039]  =======================
> [   20.180039] Code:  Bad EIP value.
> 
> 	Ingo

Thanks.
arch_setup_additional_pages -> map_compat_vdso -> __set_fixmap -> native_set_fixmap

How about this one?
---
From: Marcin Slusarz <marcin.slusarz@gmail.com>
Subject: [PATCH v2] x86: fix section mismatch warning - spp_getpage()

WARNING: vmlinux.o(.text+0x17a3e): Section mismatch in reference from the function set_pte_vaddr_pud() to the function .init.text:spp_getpage()
The function set_pte_vaddr_pud() references
the function __init spp_getpage().
This is often because set_pte_vaddr_pud lacks a __init
annotation or the annotation of spp_getpage is wrong.

spp_getpage is called from __init (__init_extra_mapping) and
non __init (set_pte_vaddr_pud) functions, so it can't be __init.
Unfortunately it calls alloc_bootmem_pages which is __init,
but does it only when bootmem allocator is available (after_bootmem == 0).

Signed-off-by: Marcin Slusarz <marcin.slusarz@gmail.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: H. Peter Anvin <hpa@zytor.com>
---
 arch/x86/mm/init_64.c |    6 +++++-
 1 files changed, 5 insertions(+), 1 deletions(-)

diff --git a/arch/x86/mm/init_64.c b/arch/x86/mm/init_64.c
index 129618c..f0923ce 100644
--- a/arch/x86/mm/init_64.c
+++ b/arch/x86/mm/init_64.c
@@ -88,7 +88,11 @@ early_param("gbpages", parse_direct_gbpages_on);
 
 int after_bootmem;
 
-static __init void *spp_getpage(void)
+/*
+ * NOTE: This function is marked __ref because it calls __init function
+ * (alloc_bootmem_pages). It's safe to do it ONLY when after_bootmem == 0.
+ */
+static __ref void *spp_getpage(void)
 {
 	void *ptr;
 
-- 
1.5.4.5

Re: [PATCH v2] x86: fix section mismatch warning - spp_getpage()

[Ingo Molnar]

* Marcin Slusarz <marcin.slusarz@gmail.com> wrote:

> -static __init void *spp_getpage(void)
> +/*
> + * NOTE: This function is marked __ref because it calls __init function
> + * (alloc_bootmem_pages). It's safe to do it ONLY when after_bootmem == 0.
> + */
> +static __ref void *spp_getpage(void)

perfect - applied to tip/x86/urgent, thanks Marcin.

	Ingo