Development around kexec ()

Neil Horman | 3 Mar 2008 15:30

[PATCH] reduce alignment constraint on ia64 elf core header

Hey all-
	Patch to reduce the alignment constraints on ia64 elf core headers.
>From what I can see there is no need to align the elf core header structure to
the EFI_PAGE_SIZE (given that the EFI will never read this data, only the kernel
will when booting a kdump kernel).  And since this is going into our reserved
memory, it saves us 3K (4K vs 1K alignment).  Tested successfully by myself on
several ia64 boxes.

Thanks & Regards
Neil

Signed-off-by: Neil Horman <nhorman@...>

 crashdump-ia64.c |    2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kexec/arch/ia64/crashdump-ia64.c b/kexec/arch/ia64/crashdump-ia64.c
index e365b3e..2aa2b12 100644
--- a/kexec/arch/ia64/crashdump-ia64.c
+++ b/kexec/arch/ia64/crashdump-ia64.c
 <at>  <at>  -231,7 +231,7  <at>  <at>  int load_crashdump_segments(struct kexec_info *info, struct mem_ehdr *ehdr,
 						       crash_memory_range,
 						       nr_ranges,
 						       &tmp, &sz,
-						       EFI_PAGE_SIZE) < 0)
+						       ELF_CORE_HEADER_ALIGN) < 0)
 				return -1;

 			elfcorehdr = add_buffer(info, tmp, sz, sz,
--

--

(Continue reading)

Bernhard Walle | 3 Mar 2008 16:56

Picon

Re: [PATCH] reduce alignment constraint on ia64 elf core header

* Neil Horman <nhorman@...> [2008-03-03 15:30]:
> Hey all-
> 	Patch to reduce the alignment constraints on ia64 elf core headers.
> >From what I can see there is no need to align the elf core header structure to
> the EFI_PAGE_SIZE (given that the EFI will never read this data, only the kernel
> will when booting a kdump kernel).  And since this is going into our reserved
> memory, it saves us 3K (4K vs 1K alignment).  Tested successfully by myself on
> several ia64 boxes.

This is wrong. Please read this thead:

https://lists.linux-foundation.org/pipermail/fastboot/2007-February/012882.html

        Bernhard

Simon Horman | 4 Mar 2008 03:52

Picon

Re: [PATCH] ia64 kern_vaddr_start was calculated incorrectly

On Mon, Mar 03, 2008 at 05:53:26PM -0800, Jay Lan wrote:
> I tested in a rhel5.1 root with:
>    2.6.24 kernel
>    kexec-tools-testing-20080227
>    crash-4.0-5.1
> 
> Crash failed to initialize:
> 
> crash: read error: kernel virtual address: a0000001007f0868  type:
> "kernel_config_data"
> WARNING: cannot read kernel_config_data
> crash: read error: kernel virtual address: a000000100f370b0  type: "xtime"
> 
> It turned out that the kexec sets info->kern_vaddr_start incorrectly
> on ia64 platform.
> 
> Jonathan Lim posted a fix to the fastboot list in March 2007:
> https://lists.linux-foundation.org/pipermail/fastboot/2007-March/013645.html
> This patch derived from Jonathan's patch, but fixes the problem at
> kexec/arch/ia64/crashdump-ia64.c.
> 
> 
> Signed-off-by: Jay Lan  <jlan@...>

Thanks Jay, Thanks Jonathan.

Sorry for not picking this up earlier. I have applied it now.

--

-- 
Horms

(Continue reading)

Simon Horman | 4 Mar 2008 05:28

Picon

Re: [PATCH] reduce alignment constraint on ia64 elf core header

On Mon, Mar 03, 2008 at 04:56:44PM +0100, Bernhard Walle wrote:
> * Neil Horman <nhorman@...> [2008-03-03 15:30]:
> > Hey all-
> > 	Patch to reduce the alignment constraints on ia64 elf core headers.
> > >From what I can see there is no need to align the elf core header structure to
> > the EFI_PAGE_SIZE (given that the EFI will never read this data, only the kernel
> > will when booting a kdump kernel).  And since this is going into our reserved
> > memory, it saves us 3K (4K vs 1K alignment).  Tested successfully by myself on
> > several ia64 boxes.
> 
> This is wrong. Please read this thead:
> 
> https://lists.linux-foundation.org/pipermail/fastboot/2007-February/012882.html

Hi Neil, Hi Bernhard,

I can confirm that with Neil's patch I end up with a zero
byte long /proc/vmcore

--

-- 
Horms

Neil Horman | 4 Mar 2008 12:49

Re: [PATCH] reduce alignment constraint on ia64 elf core header

On Tue, Mar 04, 2008 at 01:28:16PM +0900, Simon Horman wrote:
> On Mon, Mar 03, 2008 at 04:56:44PM +0100, Bernhard Walle wrote:
> > * Neil Horman <nhorman@...> [2008-03-03 15:30]:
> > > Hey all-
> > > 	Patch to reduce the alignment constraints on ia64 elf core headers.
> > > >From what I can see there is no need to align the elf core header structure to
> > > the EFI_PAGE_SIZE (given that the EFI will never read this data, only the kernel
> > > will when booting a kdump kernel).  And since this is going into our reserved
> > > memory, it saves us 3K (4K vs 1K alignment).  Tested successfully by myself on
> > > several ia64 boxes.
> > 
> > This is wrong. Please read this thead:
> > 
> > https://lists.linux-foundation.org/pipermail/fastboot/2007-February/012882.html
> 
> Hi Neil, Hi Bernhard,
> 
> I can confirm that with Neil's patch I end up with a zero
> byte long /proc/vmcore
> 
> -- 
> Horms

Rescinded.  My bad, I didn't think to check that.

Thanks!
Nei

--

-- 
/****************************************************

(Continue reading)

Simon Horman | 5 Mar 2008 02:57

Picon

Re: [PATCH] reduce alignment constraint on ia64 elf core header

On Tue, Mar 04, 2008 at 06:49:44AM -0500, Neil Horman wrote:
> On Tue, Mar 04, 2008 at 01:28:16PM +0900, Simon Horman wrote:
> > On Mon, Mar 03, 2008 at 04:56:44PM +0100, Bernhard Walle wrote:
> > > * Neil Horman <nhorman@...> [2008-03-03 15:30]:
> > > > Hey all-
> > > > 	Patch to reduce the alignment constraints on ia64 elf core headers.
> > > > >From what I can see there is no need to align the elf core header structure to
> > > > the EFI_PAGE_SIZE (given that the EFI will never read this data, only the kernel
> > > > will when booting a kdump kernel).  And since this is going into our reserved
> > > > memory, it saves us 3K (4K vs 1K alignment).  Tested successfully by myself on
> > > > several ia64 boxes.
> > > 
> > > This is wrong. Please read this thead:
> > > 
> > > https://lists.linux-foundation.org/pipermail/fastboot/2007-February/012882.html
> > 
> > Hi Neil, Hi Bernhard,
> > 
> > I can confirm that with Neil's patch I end up with a zero
> > byte long /proc/vmcore
> > 
> > -- 
> > Horms
> 
> Rescinded.  My bad, I didn't think to check that.

Thanks to Bernhard for spotting this,
I likely would have merged it otherwise.

--

--

(Continue reading)

Scott D. Davilla | 5 Mar 2008 04:02

bug/patch for i386 EFI boot

Problem:
	kexec fails under i386 EFI boot for hardware that does not 
have a VGA bios or PC bios present. The EFI bootloader properly set 
the initial screen_info boot params but kexec assumes an 
orig_video_isVGA type of 1. This causes the kexec'd kernel assume the 
presence of both VGA bios and normal PC BIOS and starts probing which 
then panics somewhere I can't see because I have no console output 
nor serial ports.

Solution:
	For EFI, efifb is a linear VGA framebuffer and can handle the 
kexec kernel switch, kexec just has to setup the kernel boot params 
properly. The below patch fixes this issue and allow a i386 EFI 
booted kernel with no VGA bios/PC bios to kexec to another kernel.

Questions:
	Why is kexec assuming i386 x86 == VGA bios in the first 
place. The existing kernel has a perfectly good screen_info stucture 
that is an exported symbol by the kernel and could be accessed by 
kexec. Since kexec is just pretending to be a boot loader, why does 
it ignore information that the real bootloader (which knows a lot 
more about the hardware) has passed? The below patch just hacks in a 
trap for efifb and will fail in the future as other x86 hardware 
without VGA bios/PC bios become available.

My first mainline submitted patch so be gentle :).

Scott

diff -Naur

(Continue reading)

Simon Horman | 5 Mar 2008 10:27

Picon

Re: bug/patch for i386 EFI boot

On Tue, Mar 04, 2008 at 10:02:31PM -0500, Scott D. Davilla wrote:
> Problem:
> 	kexec fails under i386 EFI boot for hardware that does not 
> have a VGA bios or PC bios present. The EFI bootloader properly set 
> the initial screen_info boot params but kexec assumes an 
> orig_video_isVGA type of 1. This causes the kexec'd kernel assume the 
> presence of both VGA bios and normal PC BIOS and starts probing which 
> then panics somewhere I can't see because I have no console output 
> nor serial ports.
> 
> Solution:
> 	For EFI, efifb is a linear VGA framebuffer and can handle the 
> kexec kernel switch, kexec just has to setup the kernel boot params 
> properly. The below patch fixes this issue and allow a i386 EFI 
> booted kernel with no VGA bios/PC bios to kexec to another kernel.
> 
> Questions:
> 	Why is kexec assuming i386 x86 == VGA bios in the first 
> place. The existing kernel has a perfectly good screen_info stucture 
> that is an exported symbol by the kernel and could be accessed by 
> kexec. Since kexec is just pretending to be a boot loader, why does 
> it ignore information that the real bootloader (which knows a lot 
> more about the hardware) has passed? The below patch just hacks in a 
> trap for efifb and will fail in the future as other x86 hardware 
> without VGA bios/PC bios become available.
> 
> My first mainline submitted patch so be gentle :).

Hi Scott,

(Continue reading)

Scott D. Davilla | 5 Mar 2008 15:57

[PATCH] kexec: fix i386 EFI boot using efifb

Problem:
	kexec fails under i386 EFI boot for hardware that does not 
have a VGA bios or PC bios present. The EFI bootloader properly set 
the initial screen_info boot params but kexec assumes an 
orig_video_isVGA type of 1. This causes the kexec'd kernel assume the 
presence of both VGA bios and normal PC BIOS and starts probing which 
then panics somewhere I can't see because I have no console output 
nor serial ports.

Solution:
	For EFI, efifb is a linear VGA framebuffer and can handle the 
kexec kernel switch, kexec just has to setup the kernel boot params 
properly. The below patch fixes this issue and allow a i386 EFI 
booted kernel with no VGA bios/PC bios to kexec to another kernel.

Questions:
	Why is kexec assuming i386 x86 == VGA bios in the first 
place. The existing kernel has a perfectly good screen_info stucture 
that is an exported symbol by the kernel and could be accessed by 
kexec. Since kexec is just pretending to be a boot loader, why does 
it ignore information that the real bootloader (which knows a lot 
more about the hardware) has passed? The below patch just hacks in a 
trap for efifb and will fail in the future as other x86 hardware 
without VGA bios/PC bios become available.

diff -rup 
kexec-tools-testing-20080227-org/kexec/arch/i386/x86-linux-setup.c 
kexec-tools-testing-20080227/kexec/arch/i386/x86-linux-setup.c
--- kexec-tools-testing-20080227-org/kexec/arch/i386/x86-linux-setup.c
+++ kexec-tools-testing-20080227/kexec/arch/i386/x86-linux-setup.c

(Continue reading)

Scott D. Davilla | 5 Mar 2008 16:08

Re: bug/patch for i386 EFI boot

>this looks good to me, though there is no need
>to comment out unneeded lines, just remove them.
>
>Also, if you could provide a sign-off line, as
>per Section 5 of http://linux.yyz.us/patch-format.html
>that would be great.
>
>Finally, your mail reader (or something else) seems
>to have mangled the patch such that leading <space><tab>
>has been turned into <space><space><tab>. This isn't a bit
>deal as its is easy enough for me to un-mangle it by hand.
>But just so you know.
>
>Thanks
>
>--
>Horms

Done and resubmitted with a proper subject line with commented out 
lines removed. VMware was mangling the leading tabs on the drag and 
drop from Linux to OS X ???

And as follow up question. It there any access to 
screen_info.orig_video_isVGA besides linking to the kernel? If there 
is access to orig_video_isVGA then kexec can setup the screen boot 
params as the bootloader intended instead of assuming a default VGA 
config. The orig_video_isVGA is the only parameter missing to clone 
the initial screen_info information.

Scott

(Continue reading)

Group

gmane.linux.kernel.kexec.

Search Archive

Page

1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 16

Articles in period: 150

March 2008

Language

Change language

Options

Current view: Threads only / Showing only 20 lines / Not hiding cited text.
Change to All messages, whole messages, or hide cited text.

Post a message
NNTP Newsgroup
Classic Gmane web interface
XML

XML

RSS Feed
List Information

About Gmane

Recent entries

Different minor kexec cleanups (22 May 2013)
[PATCH] makedumpfile: buddy identification when noncyclic (15 May 2013)
[PATCH] makedumpfile: message on non-mmap kernels (15 May 2013)
[PATCH] makedumpfile: non-cyclic default (15 May 2013)
[PATCH] makedumpfile: turn off mmap for free page scan (15 May 2013)
[PATCH] makedumpfile: warn on vmlinux without dwarf (15 May 2013)
[PATCH] kexec: check if memory is reserved only when loading kdump ker (14 May 2013)
[PATCH] kexec: Replace printf() with die() to error out to stderr (14 May 2013)
AFFAIRES ET PERSONNEL PRÊT OFFRE. (14 May 2013)
Kernel panics when using kexec for rebooting (13 May 2013)
[PATCH]Add kmsg_dump() to kexec path (9 May 2013)
crash read error: kernel virtual address / vmcore address mismatch ? (7 May 2013)
kernel parameter 'memmap ' is failing to ioremap with error "Malformed (6 May 2013)
[PATCH 0/4] kdump: Allow ELF header creation in new kernel (6 May 2013)
Fwd: Unknown rel relocation: Error (1 May 2013)
[PATCH] kexec: Parse percpu note size from kernel (30 Apr 2013)
[PATCH] kexec: cleanup: make add_backup_segments static (27 Apr 2013)
Investment (24 Apr 2013)
Why kdump just support ELF64CLASS? (23 Apr 2013)
/proc/vmcore kernel patches (23 Apr 2013)
makedumpfile parallel dumping test (22 Apr 2013)

Archives

50	May 2013
203	April 2013
303	March 2013
144	February 2013
186	January 2013
124	December 2012
311	November 2012
112	October 2012
104	September 2012
133	August 2012
64	July 2012
88	June 2012
55	May 2012
70	April 2012
54	March 2012
88	February 2012
65	January 2012
47	December 2011
91	November 2011
75	October 2011
89	September 2011
99	August 2011
101	July 2011
25	June 2011
56	May 2011
51	April 2011
61	March 2011
37	February 2011
36	January 2011
33	December 2010
57	November 2010
30	October 2010
99	September 2010
95	August 2010
118	July 2010
52	June 2010
89	May 2010
106	April 2010
68	March 2010
88	February 2010
39	January 2010
21	December 2009
74	November 2009
58	October 2009
24	September 2009
42	August 2009
43	July 2009
73	June 2009
55	May 2009
26	April 2009
15	March 2009
43	February 2009
93	January 2009
21	December 2008
193	November 2008
235	October 2008
141	September 2008
222	August 2008
282	July 2008
203	June 2008
171	May 2008
104	April 2008
150	March 2008
100	February 2008
85	January 2008
123	December 2007
97	November 2007
154	October 2007
220	September 2007
150	August 2007
156	July 2007
47	June 2007

Design

Zawodny | Right Menu | Left Menu

Your Own Design

Paste the URL of your CSS below. Download CSS template