| Age | Commit message (Collapse) | Author |
|---|
|
libxc from Xen 4.4 added xc_kexec_load() which will be required to
load images into Xen in the future.
Remove all the #ifdef'ery for older versions of libxc.
Signed-off-by: David Vrabel <david.vrabel@citrix.com>
Reviewed-by: Daniel Kiper <daniel.kiper@oracle.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
This patch provides support for the new Power PC litte endian (LE) mode. The
LE mode only differs in the way the instructions and data are stored in memory
thus there is no real need to duplicate the ppc64 code.
However some compilation's options, especially for the purgatory, differ
between little and big endian mode's support. A new "SUBARCH" build variable
is introduced which is currently only used for PPC64 to specify the
endianness.
Another set of changes in this patch is fixing minor endianess issues in the
ppc64 code and fix an alignment issue raised on Power7 little endian mode.
Among these fixes, the check on the kernel binary endianess is removed,
since we can imagine kexecing a LE kernel from a BE environment, as far as
the specified root filesystem and initrd file are containing the right
binaries.
This patch depends on the patch "kexec/ppc64: use common architecture
fs2dt.c file" I sent earlier on the kexec mailing list.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Following the commit 'b3c2962 fs2dt: Add a generic copy of fs2dt' which
introduced a generic fs2dt file, this patch is removing the ppc64
architecture's one.
Tests have been done successfully on Power 7 plateform.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Currently, kexec on arm assumes that it's safe to place binary images
such as atags, dtb or initrd at an estimated offset after the load
address. That estimated offset is set to 4 times the size of the
compressed image, hence assuming a minimum compression ratio of 1:4.
While that assumption matches what the in-kernel compressors are able to
achive, it doesn't take into account the .bss section the kernel image
carries, and which can grow to arbitrary sizes while not accounting to
the compressed image size.
After decompression, and before the execution of the compressed kernel,
the .bss area is initialized to zeros, trampeling over the binary images
in case they happen to live in that piece of memory.
Unfortunately, determining the full image size is not easiliy possible
at runtime, as it would include doing all possible ways of
decompression and then walk the ELF sections by hand.
For now, allow users to override the static offset with a new, arm
specific command line argument. Users are supposed to set this, and
determine a sane value by using 'arm-linux-size vmlinux'.
Signed-off-by: Daniel Mack <zonque@gmail.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
While reviewing fs2dt.c in the common kexec directory, in order to use it to
support little endian ppc64 architecture, I found some endianess
conversion's issues.
In dt_reserve, dt_base is a pointer and thus should not be converted.
In checkprop, values read from the device tree are big endian encoded and
should be converted if CPU is running in little endian mode.
In add_dyn_reconf_usable_mem_property__, fix extraneous endianess conversion
of rlen which should be natively used to increase the dt pointer.
Signed-off-by: Laurent Dufour <ldufour@linux.vnet.ibm.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
The putnode() routine in fs2dt.c was not checking for errors
returned from calls to read(). Add checks for these errors
and abort the setup of printing from purgatory if the checks
fail.
Signed-off-by: Geoff Levand <geoff@infradead.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Add a local variable 'result' to the putnode() routine of
ds2dt and use it to hold return values of calls to read().
Fixes build warnings like these:
kexec/fs2dt.c: warning: ignoring return value of ‘read’
Signed-off-by: Geoff Levand <geoff@infradead.org> for Huawei, Linaro
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
When calling locate_hole() with "hole_size" equal to the size of an
available memory block, it fails to use that memory block.
"end" and "hole_max" point to the last byte within the range, hence
- "size = end - start" is one less than "hole_size",
- "hole_base + hole_size" is one more than "hole_max".
Subtract one from "hole_size" when doing the comparison (adding 1 to "size"
could overflow in case of one big range covering the whole address space).
But explicitly check if "hole_size" is zero first, to handle this case
without causing underflows.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Initialize a local variable to zero.
Fixes build warnings like these:
kexec/kexec-elf-rel.c: warning: ‘rel.r_offset’ may be used uninitialized in this function
Signed-off-by: Geoff Levand <geoff@infradead.org> for Huawei, Linaro
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
The main kexec option code handles the -? option. Remove all the
duplicate -? handlers in the arch code which are never used.
Signed-off-by: Geoff Levand <geoff@infradead.org> for Huawei, Linaro
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Add '?' to the short option string. Fixes runtime errors like
these:
kexec: invalid option -- '?'
Signed-off-by: Geoff Levand <geoff@infradead.org> for Huawei, Linaro
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
We should check if the initrd is in uImage format, when
the kernel might be in ELF.
Signed-off-by : Suzuki K Poulose <suzuki@in.ibm.com>
Signed-off-by : Athira Rajeev<atrajeev@in.ibm.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
If the primary kernel doesn't use an initrd,
we may not have linux,initrd-* entries in the
device-tree, and hence the initial flat tree
may not contain them.
Make sure we add the entries in the dtb if the
second kernel needs an initrd.
Signed-off-by: Suzuki K. Poulose <suzuki@in.ibm.com>
Signed-off-by: Athira Rajeev<atrajeev@in.ibm.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
It's confusing that some pointers are printed as hex numbers prefixed with
0x, while some other values are printed as hex numbers without prefixes.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Cris doesn't have support for crash kernels yet.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Presumably these had been copied from ppc.
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
"kexec -p" fails to load kernels with version of the form x.y instead of
x.y.z with an error message similar to "Unsupported utsname.release:
3.10-1-amd64". Code in kernel_version() also checks the wrong variable
name for error return value from strtoul() for "minor" and "patch", and
hence possibly missing a real error.
These changes fix both of these problems.
Signed-off-by: Khalid Aziz <khalid@gonehiking.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Prefix local include paths with $(srcdir)/.
Fixes build errors like these when building for ARM out of the source tree:
cc1: fatal error: kexec/arch/arm/crashdump-arm.h: No such file or directory
Signed-off-by: Geoff Levand <geoff@infradead.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
saved_max_mem is used to calculate the amount of memory that the previous
kernel used. It seems in sh, we just calculate this variable, but we
never use it. So remove it.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
saved_max_mem is used to calculate the amount of memory that the previous
kernel used. And passed to the dump-capture kernel by kernel commandline
parameter "savemaxmem=". But in the dump-capture kernel, we never use
this parameter now, so remove saved_max_mem and don't add "savemaxmem="
to new kernel commandline.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
saved_max_mem is used to calculate the amount of memory that the previous
kernel used. And passed to the dump-capture kernel by kernel commandline
parameter "savemaxmem=". But in the dump-capture kernel, we never use
this parameter now, so remove saved_max_mem and don't add "savemaxmem="
to new kernel commandline.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
saved_max_mem is used to calculate the amount of memory that the previous
kernel used. And passed to the dump-capture kernel by kernel commandline
parameter "savemaxmem=". But in the dump-capture kernel, we never use
this parameter now, so remove saved_max_mem and don't add "savemaxmem="
to new kernel commandline.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Acked-by: "Eric W. Biederman" <ebiederm@xmission.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
several options are missed in man page, add it now. They are
--reuseinitrd, --load-preserve-context, --load-jump-back-helper
and --entry=<addr>.
Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
"--entry" option is used to specify jump back entry address, but
lack description in help message and man page. Now add it.
Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
In commit 28d4ab53, the arch-specific --debug are moved to
generic place, but the relevant description was not updated
accordingly in man page. Here change it.
Signed-off-by: Baoquan He <bhe@redhat.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
The return will never be reached, so remove it.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
There is no elf.h in this directory, so remove the useless
include.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
When borrowing codes from ia64 architecture, this comment was
forgotten to be change. So fix it.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
When memory for crashkernel isn't reserved, it's confusing that kexec
spits error message for unloading kdump kernel as if we were loading it:
# kexec -p -u
Memory for crashkernel is not reserved
Please reserve memory by passing "crashkernel=X@Y" parameter to the kernel
Then try loading kdump kernel
# echo $?
1
It's more appropriate to test if memory is reserved only when
loading kdump kernel. With this patch:
# kexec -p -u
# echo $?
0
It's also the same behavior with the case of trying to unload kernel
from unloaded state.
Signed-off-by: WANG Chao <chaowang@redhat.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Error messages are mixed stderr with stdout when we use die() along with
printf(). So use die() to keep error out consistent.
Signed-off-by: WANG Chao <chaowang@redhat.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
We used 1024 as the percpu crash note size. But for new kernel that
exports the real crash note size, we should parse it instead of
using 1024.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
The function add_backup_segments is only called by my_load and
they are in the same file. So unnecessary to export it. Make it
static.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
This reverts commit e35aa29fb40b37bf86d980b2e19af5e01c2d2549.
This patch is based on the commit 49320340f705694e387d794f7f19d407ad9baefa
"kexec: lengthen the kernel command line image"
Since the latter commit has been reverted due to its useless, this
patch should be reverted too.
Besides, This patch also changed a kernel restriction of max segments
from 16 to 70. Though kexec-tools could have more segments, more than 16,
the kexec_load syscall will still fail for the kernel side has a restriction
of 16.
Cc: Cliff Wickman <cpw@sgi.com>
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Vivek found specical handling crashkernel low in not good.
We should extend kexec-tools to handle multiple Crash kernel instead.
Extend crash_reserved_mem to array instead and use
kexec_iomem_for_each_line directly. After that we can drop
crashkernel low.
-v2: fix left over calling of parse_iomem_single() found by Vivek.
Suggested-by: Vivek Goyal <vgoyal@redhat.com>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Teach uImage_probe_xxx() to return the information about
a corrupted image. This is required to prevent the loading
of a corrupted ramdisk, where we don't have strict checking
for the other formats, unlike the kernel. So, we should abort
the operation than causing a problem with the new kernel.
Without this patch, a corrupted uImage ramdisk is treated as
a plain ramdisk where there is no format check involved.
# kexec -l uImage --initrd romfs-initrd.corrupt
The data CRC does not match. Computed: 867e73f7 expected 8f097cc0
# echo $?
0
# kexec -e
Starting new kernel
Bye!
Reserving 55MB of memory at 70MB for crashkernel (System RAM: 256MB)
Using Xilinx Virtex440 machine description
Linux version 3.6.0-rc3 (root@suzukikp) (gcc version 4.3.4 [gcc-4_3-branch revision 152973] (GCC) ) #66 Tue Apr 16 06:36:56 UTC 2013
Found initrd at 0xcf5f8000:0xcfff8040
...
NET: Registered protocol family 17
RAMDISK: Couldn't find valid RAM disk image starting at 0.
List of all partitions:
No filesystem could mount root, tried: ext2 cramfs
Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(1,0)
With this patch :
# kexec -l uImage --initrd romfs-initrd.corrupt
uImage: The data CRC does not match. Computed: 867e73f7 expected 8f097cc0
uImage: Corrupted ramdisk file romfs-initrd
With a corrupted kernel image(the behaviour remains the same) :
# kexec -l uImage.corrupt --initrd romfs-initrd
uImage: The data CRC does not match. Computed: 285787b7 expected e37f65ad
Cannot determine the file type of uImage.corrupt
Signed-off-by: Suzuki K. Poulose <suzuki@in.ibm.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
We needn't get the initial values for every regs in regs32
from symbol entry32_regs for we will manually initialize them
one by one next:
/*
* Initialize the 32bit start information.
*/
regs32.eax = 0; /* unused */
regs32.ebx = 0; /* 0 == boot not AP processor start */
regs32.ecx = 0; /* unused */
regs32.edx = 0; /* unused */
regs32.esi = setup_base; /* kernel parameters */
regs32.edi = 0; /* unused */
regs32.esp = elf_rel_get_addr(&info->rhdr, "stack_end"); /* stack, unused */
regs32.ebp = 0; /* unused */
regs32.eip = kernel32_load_addr; /* kernel entry point */
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
command line instead of empty command line
This patch prevents the problems to happen below:
In setup_linux_bootloader_parameters_high
120 cmdline_ptr = ((char *)real_mode) + cmdline_offset;
121 memcpy(cmdline_ptr, cmdline, cmdline_len);
122 cmdline_ptr[cmdline_len - 1] = '\0';
if cmdline_len == 0, Line 122 will corrupt kernel16 buf just before the commandline.
And in do_bzImage_load, for example,
369 cmdline_end = setup_base + kern16_size_needed + command_line_len - 1;
370 elf_rel_set_symbol(&info->rhdr, "cmdline_end", &cmdline_end,
371 sizeof(unsigned long));
Line 369 will go wrong, too.
Signed-off-by: Wang YanQing <udknight@gmail.com>
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Enforce size check for kernel command line to make sure it
doesn't overflow COMMAND_LINE_SIZE.
Reported-by: Nathan D. Miller <nathanm2@us.ibm.com>
Signed-off-by: Suzuki K. Poulose <suzuki@in.ibm.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
All other architectures use the command line option --dtb to pass
a dtb file name. For consistency add that option to ppc64.
Signed-off-by: Geoff Levand <geoff@infradead.org>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
When compile kexec-tools in i386, we got the following warnings:
kexec/kexec-elf-rel.c: In function ‘elf_rel_set_symbol’:
kexec/kexec-elf-rel.c:517: warning: format ‘%ld’ expects type ‘long int’, but argument 4 has type ‘size_t’
kexec/kexec-elf-rel.c: In function ‘elf_rel_get_symbol’:
kexec/kexec-elf-rel.c:541: warning: format ‘%ld’ expects type ‘long int’, but argument 4 has type ‘size_t’
This is because the two functions output a size_t value as %ld
when it should be %zd, resulting in this warning.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Functions:
- cmdline_add_memmap()
- cmdline_add_memmap_acpi()
- cmdline_add_memmap_reserved()
is kind of similar, So add a new function cmdline_add_memmap_internal() to
hold the common codes, reducing the duplication.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
fatal() nearly does the same thing as die() does, so this is kind of
duplicate. Remove fatal() and use die() instead.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Use die() to simplify code.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
when specified a wrong --entry option, it outputs the error message:
"Bad option value in --load-jump-back-helper=%s\n"
which is obviously wrong, it should be:
"Bad option value in --entry=%s\n"
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
In elf_x86_64_load, allocated memory may not be free'd if the code
exits abnormally, by calling die() or return. So the patch fixes
the possible memory leak.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Since get_command_line returns dynamically allocated memory, it is
easy for the caller to forget freeing the memory. Here fixes a
memory leak caused by this function.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Since get_command_line returns dynamically allocated memory, it is
easy for the caller to forget freeing the memory. Here fixes a
memory leak caused by this function.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
Since get_command_line returns dynamically allocated memory, it is
easy for the caller to forget freeing the memory. Here fixes a
memory leak caused by this function.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
|
In ppc_load_bare_bits, allocated memory may not be free'd if the code
exits abnormally, by calling die() or return. So the patch fixes
the possible memory leak.
This patch is also a preparation for patch10.
Signed-off-by: Zhang Yanfei <zhangyanfei@cn.fujitsu.com>
Signed-off-by: Simon Horman <horms@verge.net.au>
|
