s390-tools/s390-tools-sles15sp3-dasdfmt-retry-BIODASDINFO-if-device-is-busy.patch
Nikolay Gueorguiev 6a6c0cb679 * Upgrade s390-tools to version 2.34 (jsc#PED-3223,jsc#PED-9591)
*** v2.34.0
* Changes of existing tools:
  - ap_tools/ap-check: Add support for vfio-ap dynamic configuration
  - dbginfo.sh: Update/Add additional DASD data collection
  - dumpconf: Add new parameter 'SCP_DATA' for SCSI/NVMe/ECKD dump devices
  - libutil: Make formatted meta-data configurable
  - s390-tools: Replace 'which' with built-in 'command -v'
  - zdump/dfi_elf: Support core dumps of vr-kernels
* Bug Fixes:
  - chzdev: Fix warning about failed ATTR writes by udev
  - rust/pv: Try again if first CRL-URI is invalid
  - rust/pvattest: Add short option for --arpk
  - zdump: Fix 'zgetdump -i' ioctl error on s390 formatted dump file
*** v2.33.1
* Bug Fixes:
  - s390-tools: Fix formatting and typos in README.md
  - s390-tools: Fix release string
*** v2.33.0
* Add new tools / libraries:
  - chpstat: New tool for displaying channel path statistics
  - libutil: Add output format helpers(util_fmt: JSON, JSON-SEQ, CSV, text pairs)
* Changes of existing tools / libraries:
  - chzdev: Add --is-owner to identify files created by zdev
  - dasdfmt: Change default mode to always use full-format (Note: affects ESE DASD)
  - libap: Significantly reduce delay time between file lock retries
  - pvattest: Rewrite from C to Rust
  - pvattest: Support additional data & user-data
  - rust/pv: Support for Attestation
* Bug Fixes:
  - chreipl: Improve disk type detection when running under QEMU
  - dbginfo.sh: Use POSIX option with uname
  - s390-tools: Fix missing hyphen escapes in the man page for many tools
  - zipl/src: Fix bugs in disk_get_info() reproducible in corner cases
 *** v2.32.0
* Changes of existing tools:
  - cpumf/lscpumf: add support for machine type 3932
  - genprotimg, pvattest, and pvsecret accept IBM signing key with Armonk as
    subject locality
  - zdump/zipl: Support for List-Directed dump from ECKD DASD
  - zkey: Detect FIPS mode and generate PBKDF for luksFormat according to it
* Bug Fixes:
  - dbginfo.sh: dash compatible copy sequence
  - rust/pv_core: Fix UvDeviceInfo::get() method
  - zipl/src: Fix leak of files if run with a broken configuration
  - zkey: Fix convert command to accept only keys of type CCA-AESDATA
* Revendored vendor.tar.gz
* Removed obsolete patches
  - s390-tools-sles15sp5-01-rust-pv-support-Armonk-in-IBM-signing-key-subject.patch
  - s390-tools-sles15sp6-02-genprotimg-support-Armonk-in-IBM-signing-key-subject.patch
  - s390-tools-sles15sp6-03-libpv-support-Armonk-in-IBM-signing-key-subject.patch
  - s390-tools-sles15sp6-04-pvattest-Fix-root-ca-parsing.patch
  - s390-tools-sles15sp6-genprotimg-makefile.patch

OBS-URL: https://build.opensuse.org/package/show/Base:System/s390-tools?expand=0&rev=219
2024-08-20 08:44:18 +00:00

57 lines
1.5 KiB
Diff

From 943e577440d74ad7f8787af2590c8ab4579a459b Mon Sep 17 00:00:00 2001
From: Hannes Reinecke <hare@suse.de>
Date: Thu, 5 Nov 2015 10:57:38 +0100
Subject: [PATCH] dasdfmt: retry BIODASDINFO if device is busy
Modern udev have the wonderful 'feature' to sending a 'change'
event whenever a device opened with O_RDWR is closed again.
The reasoning is that the said program _might_ have changed
the partition table and hence we _might_ have missed a partition
update.
But in doing so it not only generated tons of pointless events
but also confused the hell out of other programs.
Idiots.
References: bsc#937340
Signed-off-by: Hannes Reinecke <hare@suse.de>
---
dasdfmt/dasdfmt.c | 19 ++++++++++++++++++-
1 file changed, 18 insertions(+), 1 deletion(-)
--- a/dasdfmt/dasdfmt.c
+++ b/dasdfmt/dasdfmt.c
@@ -621,7 +621,7 @@
*/
static void check_disk(void)
{
- int err;
+ int err, index = 0 ;
bool ro;
err = dasd_is_ro(g.dev_node, &ro);
@@ -631,6 +631,23 @@
if (ro)
error("Disk %s is read only!", g.dev_path);
if (!g.force) {
+ /*
+ * udev strikes again.
+ * Modern udev will issue a 'change' event whenever
+ * a device opened with O_RDWR is closed again.
+ * On the grounds that program _might_ have changed
+ * the partition table.
+ * And confusing the hell out ouf anyone else.
+ * Bah.
+ */
+ for ( index = 0 ; index < 6 ; index++ ) {
+ if (g.dasd_info.open_count > 1) {
+ dasd_get_info(g.dev_node, &g.dasd_info);
+ sleep(1);
+ }
+ else break;
+
+ }
if (g.dasd_info.open_count > 1)
error("Disk %s is in use!", g.dev_path);
}