The current code has different paths to construct the kernel parameters
depending on whether it's a dump kernel or a normal kernel. This would
require adding special handling to get_common_component(), because for
normal kernels the parmline is already set, contrary to dump kernels,
where the command line is constructed later.
To make the code simpler and fix a bug where the default command line no
longer works, move the dump command line processing to an earlier stage.
Also rename the old function to make the function name match what it is
actually doing.
Fixes: 5fb6434548 ("zipl: add get_common_components() and finalize_common_address_data()")
Signed-off-by: Sven Schnelle <svens@linux.ibm.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
This avoids exporting explicit settings if it's the default anyway.
I.e. avoid the last line in the example below:
$ chzdev --export - --active --type --all
[active zfcp-lun 0.0.1941:0x500507630904d...:0x4055408f00000000]
scsi_dev/queue_depth=32
Suggested-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Currently, each generated loader binary is linked into one data.o object
file which is then linked against boot.o. The data.o file is also used
to generate the data.h header file using nm, which will contain all
symbols required by zipl.
The linking process is somewhat depended on the directory, as the symbol
names are derived from the file names of the input files, which is the
default behavior of the linker. The file name in this case being the
entire path.
Examples for the current situation (starting from root dir in each case):
$ cd zipl/boot && make eckd0_cdl.bin
$ cc -no-pie -static -nostdlib -Wl,--relocatable -Wl,--format,binary -o data.o eckd0_cdl.bin
$ nm data.o
0000000000000018 D _binary_eckd0_cdl_bin_end
0000000000000018 A _binary_eckd0_cdl_bin_size
0000000000000000 D _binary_eckd0_cdl_bin_start
$ make -C zipl/boot eckd0_cdl.bin
$ cc -no-pie -static -nostdlib -Wl,--relocatable -Wl,--format,binary -o data.o zipl/boot/eckd0_cdl.bin
$ nm data.o
000000000000018 D _binary_zipl_boot_eckd0_cdl_bin_end
000000000000018 A _binary_zipl_boot_eckd0_cdl_bin_size
000000000000000 D _binary_zipl_boot_eckd0_cdl_bin_start
The example above shows that the entire path would end up in the symbol
name if specified that way.
To make this more robust, future proof, and get more control of the
resulting symbol names, use '.incbin' and some macros for embedding the
loader binary files. This also reduces the linker step and the
generation of the data.h header file.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Co-developed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Use a linker script instead of a linker flags for the layout. While at
it, use common naming `_start` for entry. This change allows us to
simplify the Makefile by using the linker option `-T` for all
bootloaders.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The stage3b loader memory layout must not be larger than 0x10000 bytes.
Let's check this in the linker script.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
+ the input section names `.text.init` and `.text.start` aren't
ambiguous, therefore there is no reason so specify a file name
+ discard `.interp` section since no loader is used
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Instead of maintaining an include list, use an exclude list (if
neccessary). This simplifies the `objcopy` calls a lot, and it's
possible because `objcopy -O binary` only includes the memory dump of
the contents of the input object file. [1] This means that sections like
for debug information that do not occupy memory during the process
execution are discarded by design. [2]
In addition, if we do not want a particular ELF section, we can list it
in the discard section of the linker script.
[1] https://man7.org/linux/man-pages/man1/objcopy.1.html
[2] https://man7.org/linux/man-pages/man5/elf.5.html
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
This disables the following warning `ld: warning: stag3a.elf has a LOAD
segment with RWX permissions` for newer linker. This changes nothing in
functionality for the bootloaders, since only the binaries and not the
ELF files are actually used.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Normally, `gcc` takes care of adding the `GNU_STACK` marking except for
assembly code. Therefore, let's add the marking manually. Discard the
`.note.GNU-stack` section in the linker script since it's just a
"message" from the compiler to the linker. This fixes the linker
warning:
ld: warning: entry.o: missing .note.GNU-stack section implies executable stack
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Use linker script for all bootloaders instead of having the load
addresses of the bootloaders hard-coded in the Makefile. This allows us
to simplify the Makefile and it fixes the dependency tracking for the
bootloaders *0.bin, *1.bin, and *1b.bin.
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
This allows the reuse of the macro in the next patch. In order to match
the naming scheme rename `STAGE1B_LOAD_ADDR` to `STAGE1B_LOAD_ADDRESS`.
Acked-by: Eduard Shishkin <edward6@linux.ibm.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Instead of maintaining an include list, use an exclude list (if
neccessary). This simplifies the `objcopy` calls a lot, and it's
possible because `objcopy -O binary` only includes the memory dump of
the contents of the input object file. [1] This means that sections like
for debug information that do not occupy memory during the process
execution are discarded by design. [2]
In addition, if we do not want a particular ELF section, we can list it
in the discard section of the linker script.
[1] https://man7.org/linux/man-pages/man1/objcopy.1.html
[2] https://man7.org/linux/man-pages/man5/elf.5.html
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
This disables the following warning `ld: warning: stage3.exec has a LOAD
segment with RWX permissions` for newer linker. This changes nothing in
functionality for the zipl bootloaders, since only the binaries and not
the ELF files are actually used.
Acked-by: Eduard Shishkin <edward6@linux.ibm.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Normally, `gcc` takes care of adding the `GNU_STACK` marking except for
assembly code. Therefore, let's add the marking manually. Discard the
`.note.GNU-stack` section in the linker script since it's just a
"message" from the compiler to the linker. This fixes the linker
warning:
ld: warning: entry.o: missing .note.GNU-stack section implies executable stack
Acked-by: Eduard Shishkin <edward6@linux.ibm.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The ELF section `.eh_frame` is used for frame unwinding during, for
example, exception handling and the section `.interp` is used by the
program interpreter. [1] Since no frame unwinding nor a program
interpreter is used for the bootloaders we can discard these input
section.
The content of the (exec|bin) files don't change after this change.
[1] https://refspecs.linuxfoundation.org/LSB_1.2.0/gLSB/specialsections.html
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add support to chreipl/lsreipl to show and reconfigure ipl parameters
for list-directed IPL from ECKD DASD.
Parameters accepted:
device - ccw device id
bootprog - Boot Prog Selector
loadparm - Load parameters
brchr - location of bootrecord in cylinder/head/record format.
clear - clear memory before loading
Signed-off-by: Sven Schnelle <svens@linux.ibm.com>
Reviewed-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
With parameter OSSL_ASYM_CIPHER_PARAM_IMPLICIT_REJECTION one can enable or
disable the implicit rejection mechanism for RSA PKCS#1 v1.5 decryption.
For the default provider, this is enabled by default.
We do not support implicit rejection so far, so disable it for the
default provider operation context when the key contains an RSA secure key,
and reject to set it to anything else then 0 (disabled).
See OpenSSL commit 5ab3ec1bb1
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Since OpenSSL 3.1 a new RSA-PSS salt length constant exists to select
the maximum possible salt length based on the RSA-PSS parameters and the
digest used: OSSL_PKEY_RSA_PSS_SALT_LEN_AUTO_DIGEST_MAX
This is the default salt length when no other salt length is set by
the caller.
In contrast to OSSL_PKEY_RSA_PSS_SALT_LEN_MAX, OSSL_PKEY_RSA_PSS_SALT_LEN_AUTO_DIGEST_MAX
also ensures that the resulting salt length is not larger than the used
digest size. The salt length calculated with OSSL_PKEY_RSA_PSS_SALT_LEN_MAX
may be larger than the digest size, dependent on the RSA-PSS parameters.
FIPS 186-4 section 5 "The RSA Digital Signature Algorithm", subsection
5.5 "PKCS #1" says: "For RSASSA-PSS […] the length (in bytes) of the
salt (sLen) shall satisfy 0 <= sLen <= hLen, where hLen is the length of
the hash function output block (in bytes)."
See OpenSSL commit 6c73ca4a2f
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
With parameter OSSL_SIGNATURE_PARAM_NONCE_TYPE one can choose to generate
a deterministic signature, i.e. using a deterministic nonce K.
We do not support that so far, so reject to set the nonce type to
anything other than 0 (random nonce K).
See OpenSSL commit f3090fc710
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Fix the improper order of resource releasing resulted in failed umount
Do umount after closing (not before)
Fixes: f7d2339 (zipl: List-Directed IPL from ECKD DASD)
Reported-by: Alexander Egorenkov <egorenar@linux.ibm.com>
Tested-by: Alexander Egorenkov <egorenar@linux.ibm.com>
Signed-off-by: Eduard Shishkin <edward6@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
The library libvmdump is written in C++ and uses exception handling
and class inheritance. The tools vmur and vmconnect are compiled using
the C++ compiler for linking with the libvmdump library.
With integration of vmdump file support into the zgetdump tool and
the removal of vmur vmdump file conversion support there is no need
anymore the vmconvert tool.
With removal of the tool vmconvert, the libvmdump library written in C++
has no more user and can be deleted.
Remove the vmconvert command from the zgetdump man page.
Remove the directories libvmdump and vmconvert and adjust the Makefile.
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Suggested-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Remove option -c for implicit dump file conversion from vmdump format
to lkcd format. If option -c is specified, an error message
is printed and the program terminates. A hint to the new conversion
tool zgetdump is shown. Option -c is not listed in the help text.
The man page is also updated to reflect this change.
With zgetdump tool now supporting vmdump file format there is no need to
convert a vmdump file while extracting it from the reader device.
Use commands
1. vmur rec <spool-id> VMDUMP_FILE
2. zgetdump VMDUMP_FILE ELF_FILE
to extract a vmdump file from the reader and convert it to elf format.
Note to Maintainer:
The rename of file vmur.cpp to vmur.c triggers the pre-commit
checking of git-format-clang. This checking fails. This is intended.
The patch shows the changes for this line item.
The pre-commit checking fails on many, many changes not touched
for changed by this patch.
The pre-commit failure should be fixed with a follow-on patch.
Suggested-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Acked-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Add support for vmdump 64big format to the zgetdump utility.
The result is an elf file suitable for tool crash.
This patch adds file dfi_vmdump.c to handle conversion from
z/VM vmdump format to the internal memory representation of this tool.
Therefore both output formats elf and s390 are supported as target
of the conversion.
The vmdump file formats 32bit and 64bit are not supported anymore.
They have been obsolete since CP version 5.1 (end of service since
30-Sep-2009).
Vector register set is not part of a vmdump file and can not be
displayed.
Output before:
# zgetdump ~/vmdump-0383
zgetdump: No valid dump found on "/root/vmdump-0383"
#
Output after:
# ./zgetdump ~/vmdump-0383 ~/vmdump-0383.elf
Format Info:
Source: vmdump
Target: elf
Copying dump:
00000000 / 00000379 MB
00000379 / 00000379 MB
Success: Dump has been copied
# file ~/vmdump-0383.elf
/root/vmdump-0383.elf: ELF 64-bit MSB core file, IBM S/390, version 1 \
.... (SYSV), SVR4-style, from 'vmlinux'
#
# ./zgetdump -i ~/vmdump-0383
General dump info:
Dump format........: vmdump
Version............: 2
UTS node name......: s8360046.lnxne.boe
UTS kernel release.: 6.1.0-rc2-d-perf+
UTS kernel version.: #1 SMP Fri Oct 28 09:42:13 CEST 2022
System arch........: s390x (64 bit)
CPU count (online).: 2
CPU count (real)...: 2
Dump memory range..: 2048 MB
Memory map:
0000000000000000 - 000000007fffffff (2048 MB)
#
Signed-off-by: Thomas Richter <tmricht@linux.ibm.com>
Reviewed-by: Mikhail Zaslonko <zaslonko@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
The `.check_dep_zgetdump` file is used to cache the result of the
dependency checks and should not be used as input for linking or
anything else. Let's add it as dependency for the objects file. This
shouldn't cause any problems since the Makefile rule for object files is
defined in `common.mak` as follows:
%.o: %.c
$(CC) $(ALL_CPPFLAGS) $(ALL_CFLAGS) -c $< -o $@
Fixes: https://github.com/ibm-s390-linux/s390-tools/issues/147
Fixes: 8d8d5e9746 ("zdump: Fix Makefile dependencies")
Signed-off-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Enforce that the first glib.h include is done via glib-helper.h for libpv
so that glib version checks are in place.
Change zdump and pvattest such that they never include glibstuff before
libpv/glib-helper.h
Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Since coreutils v9.1 commit 8f31074cb ("chown: warn about USER.GROUP")
chown utility now warns about using of wrong separator for USER and
GROUP options. This leads to warnings during intall step in ziomon:
$ make install
INSTALL ziomon
SED ziomon/install
chown: warning: '.' should be ':': ‘root.root’
SED ziomon/install
chown: warning: '.' should be ':': ‘root.root’
SED ziomon/install
chown: warning: '.' should be ':': ‘root.root’
To fix those warnings use ':' as separator instead of '.'. With this
patch applied those warnings are no longer present.
Signed-off-by: Fedor Loshakov <loshakov@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
As on space problems the packing may fail, incomplete tgz are send.
Today no evidence for this is in dbginfo.log, packed upfront into the tgz.
This update helps to identify space related corruption of tgz files.
Reviewed-by: Mario Held <mario.held@de.ibm.com>
Signed-off-by: Joern Siglen <siglen@de.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
When the kernel is build with CONFIG_NVME_MULTIPATH enabled and an NVMe
where the driver exposes its controller separately is used, chreipl node
on a mount point of the NVMe fails. The failure occurs because chreipl
tries to get the function_id from the NVMe's nvmeXnY block device via
the path '/sys/block/nvmeXnY/device/device/function_id' which does not
exist.
The underlying problem is that with NVMe multipath support the PCI
function is not associated directly with a particular NVMe namespace but
instead with the entire NVMe device which may host multiple NVMe
namespaces. The correct way to get from the block device to the
underlying PCI device and its function ID is then to extract the device
name i.e. the 'nvmeX' part from 'nvmeXnY' and use that to get the
function ID via '/sys/class/nvme/nvme<X>/device/function_id'.
Fixes: 0472b5ea5c ("ipl-tools: Add nvme device support to lsreipl/chreipl")
Reviewed-by: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Signed-off-by: Niklas Schnelle <schnelle@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
xxd has vim-common as a dependency. We can avoid that by using od
instead which is from coreutils.
While at it, add the require_command function from the
pvattest-info script to the pvextract-hdr script and use it.
Suggested-by: Christian Borntraeger <borntraeger@linux.ibm.com>
Reviewed-by: Marc Hartmayer <mhartmay@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Fix issue in the lszdev command where column names if using the --pairs
command line switch can contain characters that are not allowed to be
used as variable names in a shell environment.
Replace bad characters in column names by an underscore if the --shell
command line switch is specified along with the --pairs switch.
The additional --shell switch mimics what is already available in the
lsblk command and thus it shall be used along with the --pairs switch.
Signed-off-by: Daniel S. Haischt <modean@linux.ibm.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
The proc_sum_t struct is defined with the packed attribute and
produces the following compile warning:
mon_procd.c: In function ‘read_cpu’:
mon_procd.c:351:27: warning: taking address of packed member of ‘struct proc_sum_t’ may result in an unaligned pointer value [-Waddress-of-packed-member]
351 | cal_cpu(&cpudata, &proc_sum.cpu);
| ^~~~~~~~~~~~~
Silence the warning with -Wno-address-of-packed-member compile flag
since the struct is required to be as is.
Reviewed-by: Gerald Schaefer <gerald.schaefer@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
The struct scsi_dump_sb is defined with the __packed attribute and
produces the following compile warning in zfcpdump:
zfcpdump_part.c: In function ‘csum_update’:
zfcpdump_part.c:125:41: warning: taking address of packed member of ‘struct scsi_dump_sb’ may result in an unaligned pointer value [-Waddress-of-packed-member]
125 | dump_sb.csum_size, &dump_sb.csum)) {
| ^~~~~~~~~~~~~
However, all members are of the same size and the __packed attribute is
unnecessary. Remove the attribute to get rid of the warning.
Add assertion to ensure correct size of the struct during compile time.
Reviewed-by: Steffen Eiden <seiden@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
The 'unistd.h' header was missing. Under some circumstances the
-D_GNU_SOURCE gcc flag does not trigger including that file.
Therefore, explicitly include this file here.
Fixes: e1aec24e84 ("libutil: introduce util_lockfile")
Signed-off-by: Steffen Eiden <seiden@linux.ibm.com>
Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
mdevctl has been updated to use /usr/lib/mdevctl/scripts.d/callouts/
instead of /etc/mdevctl.d/scripts.d/callouts/. The /etc location
is considered deprecated, meaning mdevctl will also look at that
location for now but might eventually stop looking in /etc for
callout scripts.
Based on that, update the installation location for the ap-check
callout. However, because older versions of mdevctl will still
only look in /etc, let's also put a wrapper script in /etc for now
to provide backward compatibility, and plan to remove it at a
later time.
Link: df6bb57429
Fixes: https://github.com/ibm-s390-linux/s390-tools/issues/139
Signed-off-by: Matthew Rosato <mjrosato@linux.ibm.com>
Reviewed-by: Jan Hoeppner <hoeppner@linux.ibm.com>
Reviewed-by: Boris Fiuczynski <fiuczy@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Commit ae7217806b ("common.mak: remove `LINK` and `LINKXX`") changed
the LINK and LINKXX definition to ensure that the same compiler is used
for compilation and linking when specified with CC= (or CXX=
respectively) on the commandline.
However, the commit also removed the pretty print for both commands. Add
the pretty print back in and use the already defined $(CC) and $(CXX)
variables to preserve the original change.
Fixes: ae7217806b ("common.mak: remove `LINK` and `LINKXX`")
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
