zipl currently uses the FIBMAP ioctl to map blocks for the bootloader;
on XFS, if FIBMAP is requested on a reflinked file, it will fail -
and FIBMAP returns 0 in this case, which is indistinguishable from a
hole. This causes boot to fail because the file is not mapped.
We can use the FIEMAP ioctl instead, which is able to map reflinked
files. While FIEMAP is able to map entire extents at once, here we
simply use it to obtain the mapping block-by-block so that it fits
in with the current FIBMAP calls.
Fixes: https://github.com/ibm-s390-tools/s390-tools/issues/34
Closes: https://github.com/ibm-s390-tools/s390-tools/pull/36
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
Reviewed-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Tested-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
With kernel 4.19 there comes an extension to the existing
AP bus which supports alternate zcrypt drivers. For details
about this see kernel patch "s390/zcrypt: AP bus support for
alternate driver(s)". So now lszcrypt displays the driver
name in verbose mode. As some of the information displayed
by lszcrypt was based on sysfs attributes, which are only
available when the default zcrypt driver is bound to the
device, this also needed some rework. If a sysfs attribute
is not available because of an alternate driver binding
(or no driver) a question mark is printed into the field.
Together with this a slight rework of the displayed information
has been done. The two columns for pending requests and pending
replies has been merged to one pending column and the column
sizes have been adjusted.
Signed-off-by: Harald Freudenberger <freude@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
zdsfs writes output for --help and --version to stderr. This is
likely because zdsfs calls into a FUSE function for additional
usage output that also writes to stderr (prior to FUSE 3.0.0).
To be consistent with other s390-tools and GNU coding guide lines,
fix this by using stdout for zdsfs generated output, and by
redirecting stderr to stdout before calling the FUSE function.
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
tape390_display writes output for --help to stderr. To be consistent
with other s390-tools and GNU coding guide lines, fix this by using
stdout for tool-generated usage help.
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add missing long options for --help and --version in util_prg_example.
Signed-off-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Both iucvconn and iucvtty write output for --help to stderr. To be
consistent with other s390-tools and GNU coding guide lines,
fix this by using stdout for tool-generated usage help.
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
hmcdrvfs writes output for --help and --version to stderr. This is
likely because hmcdrvfs calls into a FUSE function for additional
usage output that also writes to stderr (prior to FUSE 3.0.0).
To be consistent with other s390-tools and GNU coding guide lines,
fix this by using stdout for hmcdrvfs generated output, and by
redirecting stderr to stdout before calling the FUSE function.
Signed-off-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
cmsfs-fuse writes output for --help and --version to stderr. This is
likely because cmsfs-fuse calls into a FUSE function for additional
usage output that also writes to stderr (prior to FUSE 3.0.0).
To be consistent with other s390-tools and GNU coding guide lines,
fix this by using stdout for cmsfs-fuse generated output, and by
redirecting stderr to stdout before calling the FUSE function.
Also remove an inactive call to add FUSE version output on --version to
be consistent with other FUSE-based s390-tools.
Signed-off-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Building zipl from outside the zipl/ sub-directory fails due to a
missing dependency:
# cd s390-tools
# make -C zipl/src/ zipl
boot.c:18:10: fatal error: ../boot/data.h: No such file or directory
#include "../boot/data.h"
This is because boot/data.h needs to be generated before src/boot.c can
be compiled. Fix this by adding the missing dependency to the zipl
Makefile.
Signed-off-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Assume "N/A" for non-zfcp SCSI devices, such as iSCSI or virtio-scsi-ccw,
to not erroneously access absent zfcp-specific sysfs attributes.
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Simply skip non-zfcp SCSI devices, such as iSCSI or virtio-scsi-ccw,
to not erroneously access absent zfcp-specific sysfs attributes.
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The default bash setting have nullglob disabled so if
$SCSI_DEV/scsi_generic*
aka
/sys/bus/scsi/devices/*:*:*:*/scsi_generic*
does not match anything, it leaves the glob pattern unmodified and
SG_DEV=$(basename $SG_DEV/*)
results in the literal:
*
If $SG_INQ exists, it invoked sg_inq with more than the one allowed
positional argument for a SCSI generic device node:
sg_inq /dev/*
Causing error messages and the usage of sg_inq to land in $TAPE_SERIAL.
To not have to rely on the nullglob setting, explicitly check for the
existence of $SCSI_DEV/scsi_generic before evaluating
SG_DEV=$(basename $SG_DEV/*).
Also handle availability of sg_inq but absence of scsi_generic
individually to provide the user with a hint if only sg is missing.
Suggested-by: Eric Farman <farman@linux.vnet.ibm.com>
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Cc: Eric Farman <farman@linux.vnet.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
This should be more stable than parsing the IBM lin_tape proc-fs output.
If sysfs is not available or provides no match, fall back to proc-fs.
An alternative to "ls -1d ... | head -n1" would have been to use
"shopt -s extglob; ls -d $SCSI_DEV/lin_tape/$DEV_NAME+([0-9])"
but I did not want to have to rely on extglob potentially influencing
other old code of lstape.
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
https://www.ibm.com/developerworks/linux/linux390/s390-tools-1.8.0.html
before the first upstream commit b627b8d8e1 ("Initial s390-tools-2.0.0
import") introduced SCSI tape/changer output for lstape. It used the SCSI
device serial number as lookup key to find a match in IBM lin_tape device
driver proc-fs output for a given SCSI device name. Since multiple paths
to the same tape/changer have the same serial number, multiple matches can
occur and distort the output:
% lstape
...
SCSI tape devices (found 12):
...
Generic Device Target Vendor Model Type State
sg0 IBMtape0 3 0:0:0:0 IBM ULT3580-TD6 tapedrv
running
sg1 IBMchanger0 1 2 3 0:0:0:1 IBM
3573-TL changer running
...
There are unexpected additional Device suffix numbers for each additional
path of the same tape/changer (independent of actual path failover
enablement in lin_tape). The multiple matches above cause excess arguments
to printf. Explaining the resulting output, the bash man page says:
"The format is reused as necessary to consume all of the arguments."
This also causes a wrong number of found devices.
Current lin_tape (v3.0.31) uses the printf format string
"%-8d%-12s%-25s%-16s%-16s%-10s" for fixed column proc-fs output.
This causes the following example input to lstape:
==> /proc/scsi/IBMchanger <==
lin_tape version: 3.0.31
lin_tape major number: 251
Attached Changer Devices:
Number model SN HBA SCSI FO Path
0 3573-TL 00L4U78W6497_LL0 Virtio SCSI HBA 0:0:0:1 Primary
1 3573-TL 00L4U78W6497_LL0 Virtio SCSI HBA 0:0:1:1 Alternate
2 3573-TL 00L4U78W6497_LL0 Virtio SCSI HBA 1:0:1:1 Alternate
3 3573-TL 00L4U78W6497_LL0 Virtio SCSI HBA 1:0:0:1 Alternate
==> /proc/scsi/IBMtape <==
lin_tape version: 3.0.31
lin_tape major number: 251
Attached Tape Devices:
Number model SN HBA SCSI FO Path
0 ULT3580-TD6 10WT037733 Virtio SCSI HBA 0:0:0:0 Primary
1 ULT3580-TD6 10WT037701 Virtio SCSI HBA 0:0:1:0 Primary
2 ULT3580-TD6 10WT037701 Virtio SCSI HBA 1:0:1:0 Alternate
3 ULT3580-TD6 10WT037733 Virtio SCSI HBA 1:0:0:0 Alternate
If that was reliable and stable, we could use fixed column parsing:
IBM_IDX=$(
awk 'BEGIN {
FIELDWIDTHS="8 12 25 16 16 10"
}
gensub(" ","","g",$5) == "'$SCSI_ID'"{
print gensub(" ","","g",$1)
}' /proc/scsi/$DEV_NAME
)
However, we can hardly guarantee a field value always fits into the left-
adjusted columns (printf won't cap to field width!). Also, the fixed column
format changed between lin_tape v3.0.20 and v3.0.18 from
"%-8d%-12s%-18s%-16s%-16s%-10s" with less width for the 'SN' field.
The column format also changed between v2.2.0 and v1.37.0 from
"%-8d%-12s%-18s%-32s%-10s" where the 'SCSI' field did not even exist.
Likely that was the reason for the original Serial lookup in lstape.
As can be seen in the example, at least the HBA device driver string can
contain whitespace and breaks the field numbers with tokenized parsing.
We want to match column 5 (SCSI) but above we would match against the
string 'SCSI' as part of "Virtio SCSI HBA" instead of e.g. '0:0:0:0'.
The following works currently knowing that the last two columns do not
contain white space. But it can also fail if the value of field
'HBA' does not fit into the left-adjusted columns so there is no white
space between columns 'HBA' and 'SCSI' making them collapse:
awk '$(NF-1) == "'$SCSI_ID'" { print $1 }' /proc/scsi/$DEV_NAME
Grep for the SCSI device name as word (to skip names with same substring,
such as 0:0:1:1 also matching 0:0:1:10) and cut the first field 'Number'.
If there is no SCSI column at all (and no SCSI LLDD or other column with a
name accidentally matching an existing SCSI device name), we get no match
and better bail out with the initialized "N/A" for the lstape column
"Device". If the SCSI column exists, the SCSI device names are unique and
we always get at most one match.
Signed-off-by: Steffen Maier <maier@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Reviewed-by: Jens Remus <jremus@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
When zfcpdump is built you get a compile warning
zfcpdump.c: In function ‘init_sig’:
zfcpdump.c:307:24: warning: cast between incompatible function types from ‘void (* (*)(int, siginfo_t *, void *))(int)’ {aka ‘void (* (*)(int, struct <anonymous> *, void *))(int)’} to ‘void (*)(int)’ [-Wcast-function-type]
g.sigact.sa_handler = (__sighandler_t)dump_sig_handler;
Furthermore the man pages for sigaction says
$man 2 sigaction
[...]
If SA_SIGINFO is specified in sa_flags, then sa_sigaction (instead
of sa_handler) specifies the signal-handling function for signum.
This function receives three arguments, as described below.
[...]
Because SA_SIGINFO is set, using sa_handler at all is wrong in this case.
Fix this by giving dump_sig_handler the correct return type and assign it
to sa_sigaction.
Signed-off-by: Philipp Rudo <prudo@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The zfcpdump kernel is built without module support. Furthermore the initrd
doesn't contain modprobe. So trying to load modules using modprobe in
zfcpdump userspace doesn't really make sense.
Signed-off-by: Philipp Rudo <prudo@linux.ibm.com>
Reviewed-by: Steffen Maier <maier@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Since version 198 (Feb 2013) systemd contains kernel-install, a script
managing kernel installs. This script allows execution of drop-in scripts
for customization. Add such a drop-in script to s390-tools to handle the
installation of zfcpdump kernels.
It's main purpose is to manage a link to the latest installed zfcpudump
kernel at ZFCPDUMP_IMAGE, where zipl expects to find the image to install
for a SCSI dumper.
The script supports two installation modes. One recommended by the
BootLoaderSpecs [1] to /boot/<machine-id>/<kernel-version> directories and
one directly to /boot. In the second case files are renamed during
installation to <original-name>-<kernel-version> to guarantee unique names.
Because the zfcpdump kernel is so special make the script stand-alone and
prevent any other script from being executed (exit 77) when a zfcpdump is
installed. Especially avoid functionality like creating an initrd (already
provided by s390-tools) or creating a boot entry (the zfcpdump kernel
should not be used for any other purpose than dumping).
The script requires systemd >= 203.
[1] https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec/
Signed-off-by: Philipp Rudo <prudo@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
In the zipl Makefile macros are defined with the sole purpose to define
other macros in zipl.h. Remove this unnecessary indirection.
Signed-off-by: Philipp Rudo <prudo@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The file system dumper no longer exists. So there is no need in keeping
'PART' to distinguish between the file system and the partition dumper.
Adjust the macro name to the usage in zipl by removing 'PART'.
While at it also adjust the file names the macros stand for.
Signed-off-by: Philipp Rudo <prudo@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The macro names are a mixture of the original file system dumper (FSDUMP)
and its extension to cut the file system and write directly to a partition
(PART). The original dumper no longer exists and today the feature is
called zfcpdump. Prevent confusion by renaming the macros to fit todays
usage.
Signed-off-by: Philipp Rudo <prudo@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Allow to specify a volume-type for a key. This applies to all
associated volumes. The volume type can be either 'plain' or
'luks2'. New keys created will default to 'luks2', but existing
keys that do not have a volume-type property default to 'plain'
for compatibility reasons.
The volume type 'luks2' is only available when the define
HAVE_LUKS2_SUPPORT is set in the makefile. This is set only
when libcryptsetup version 2.0.3 or newer is available
at build time. If the define is not set, the volume-type
option is not available to the user, and the volume-type of
a key defaults to 'plain'.
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Store a verification pattern in the properties file along
with the secure key. The verification pattern allows to identify
the inner key even when the secure key is no longer valid.
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The zkey-cryptsetup tool is used to validate and re-encipher
secure AES volume keys of volumes encrypted with LUKS2 and
the paes cipher.
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Reviewed-by: Hendrik Brueckner <brueckner@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The IPAPPEND option in the pxelinux config file can be used
to pass information about the boot interface to the booted
Linux kernel.
The support for this option was missing and could lead to
installation failures, e.g. using kickstart.
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Client architecture aware DHCP configurations may send a
bootfile for the wrong architecture, if the DHCP request
doesn't include the client architecture identifier 0x001f.
Further, the DHCP reply may include the pxepathprefix option,
which points to the base directory to use for TFT requests.
If the value is not honored by the client, the wrong
configuration file, kernel, etc. could be loaded.
Both issues are fixed by this commit.
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Compressed kernel modules (filename *.ko.xz) would not be included
since the script was looking for files with .ko suffix only.
This can result in a pxelinux.0 image that can't access the
network, which is not good for a network boot loader...
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The output of /proc/<pid>/stat will show the process name in parentheses.
The parsing code in read_stat() tries to filter out the parentheses, which
will go wrong when the process name itself also contains parentheses, e.g.
in an output like this: "2421 ((sd-pam)) S 2420 2420 2420 ..."
In this case, the first closing parentheses will be taken as end marker,
and the sscanf() on the remaining string will silently fail, leaving its
values in uninitialized state and producing wrong data.
Fix this by using strrchr() instead of strchr() to find the last closing
parentheses. Also add return value checking for sscanf() and initialize the
values to 0.
Signed-off-by: Gerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
mk-pxelinux-ramfs, mod_fsstatd|procd.service, zfcpdump all use
/sbin/modprobe path. Adjust zdev to use that path as well. This works
universally on both usr-merge systems (e.g. Debian/Ubuntu with
usrmerge package installed), and split-usr systems alike
(Debian/Ubuntu current stable releases default).
Closes: #31
Link: https://bugs.launchpad.net/ubuntu/+source/s390-tools/+bug/1777600
Signed-off-by: Dimitri John Ledkov <xnox@ubuntu.com>
Acked-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
When a volume or APQN association is made while generating or
importing a key, and a duplicate association is detected, then
this may leave the key in an inconsistent state.
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
When a zkey generate or change command is used to associate one
or multiple APQNs the command succeeds, but no key is generated
and no APQNs are associated, because the return code returned by
_keystore_apqn_check() is wrong.
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The APQN check routine as well as the properties helper functions
do not free all memory that they allocated.
Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
Add a zipl-switch-to-blscfg script that can be used to switch the zipl
configuration in a system to use the BootLoaderSpec (BLS) config files
to define the IPL sections instead of having them defined in zipl.conf.
Closes: #28
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
The BootLoaderSpec (BLS) defines a file format for boot configurations,
so bootloaders can parse these files and create their boot menu entries
by using the information provided by them [0].
This allow to configure the boot items as drop-in files in a directory
instead of having to parse and modify a bootloader configuration file.
If the /boot/loader/entries exists and there are BLS files there, then
these are parsed and configuration sections are added without the need
to have these in a zipl.conf file.
A different BLS directory can be specified from the command line using
the --blsdir option.
[0]: https://www.freedesktop.org/wiki/Specifications/BootLoaderSpec/
GitHub-ID: #28
Signed-off-by: Javier Martinez Canillas <javierm@redhat.com>
Reviewed-by: Stefan Haberland <sth@linux.ibm.com>
Signed-off-by: Jan Höppner <hoeppner@linux.ibm.com>
