pool/drbd
SHA256
2
0
Fork 1
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Accepting request 1113586 from home:hmzhao:branches:network:ha-clustering:Factory

Browse Source 
drbd: fix build error against kernel v6.5.4 (boo#1215699)

OBS-URL: https://build.opensuse.org/request/show/1113586
OBS-URL: https://build.opensuse.org/package/show/network:ha-clustering:Factory/drbd?expand=0&rev=146
This commit is contained in:
heming zhao 2023-09-26 05:51:01 +00:00 committed by Git OBS Bridge
parent 90d03da070
commit e7b40eeaea
23 changed files with 3004 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

37
0001-drbd-allow-transports-to-take-additional-krefs-on-a-.patch Normal file
View File

@ -0,0 +1,37 @@
From 0dda200877d1b801fba948a5948f321bee1a75a9 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Fri, 1 Sep 2023 08:03:01 +0200
Subject: [PATCH 01/20] drbd: allow transports to take additional krefs on a
listener
by making the drbd_listener_destroy() available to the transports.
heming.zhao@suse.com
- For compiling drbd, I modified this patch. please differ with upstream
to find changes.
---
drbd/drbd_transport.c | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/drbd/drbd_transport.c b/drbd/drbd_transport.c
index d1ec5724dec6..416a50499046 100644
--- a/drbd/drbd_transport.c
+++ b/drbd/drbd_transport.c
@@ -223,7 +223,7 @@ int drbd_get_listener(struct drbd_transport *transport, struct drbd_path *path,
return err;
}
-static void drbd_listener_destroy(struct kref *kref)
+void drbd_listener_destroy(struct kref *kref)
{
struct drbd_listener *listener = container_of(kref, struct drbd_listener, kref);
struct drbd_resource *resource = listener->resource;
@@ -331,3 +331,4 @@ EXPORT_SYMBOL_GPL(drbd_find_path_by_addr);
EXPORT_SYMBOL_GPL(drbd_stream_send_timed_out);
EXPORT_SYMBOL_GPL(drbd_should_abort_listening);
EXPORT_SYMBOL_GPL(drbd_path_event);
+EXPORT_SYMBOL_GPL(drbd_listener_destroy);
--
2.35.3

65
0002-drbd-improve-decision-about-marking-a-failed-disk-Ou.patch Normal file
View File

@ -0,0 +1,65 @@
From f2cd05b8d60d27f43b07175b92ef4c2a69b8e3a2 Mon Sep 17 00:00:00 2001
From: Joel Colledge <joel.colledge@linbit.com>
Date: Wed, 6 Sep 2023 15:49:44 +0200
Subject: [PATCH 02/20] drbd: improve decision about marking a failed disk
Outdated
Sometimes it is possible to update the metadata even after our disk has
failed. We were too eager to remove the MDF_WAS_UP_TO_DATE flag in this
case.
Firstly, we used the "NOW" states, so would mark our metadata Outdated
if we were a Primary with UpToDate data and no peers, and our disk
failed. Use the "NEW" states instead.
Secondly, do not consider peers that are disconnecting, because they
will not see that our disk state is Failed, and so will outdate
themselves. We do not want to outdate both nodes in this situation.
---
drbd/drbd_state.c | 18 ++++++++++++++----
1 file changed, 14 insertions(+), 4 deletions(-)
diff --git a/drbd/drbd_state.c b/drbd/drbd_state.c
index 7e6e3477893d..8b60afeb097b 100644
--- a/drbd/drbd_state.c
+++ b/drbd/drbd_state.c
@@ -2489,15 +2489,24 @@ static void initialize_resync(struct drbd_peer_device *peer_device)
/* Is there a primary with access to up to date data known */
static bool primary_and_data_present(struct drbd_device *device)
{
- bool up_to_date_data = device->disk_state[NOW] == D_UP_TO_DATE;
- bool primary = device->resource->role[NOW] == R_PRIMARY;
+ bool up_to_date_data = device->disk_state[NEW] == D_UP_TO_DATE;
+ struct drbd_resource *resource = device->resource;
+ bool primary = resource->role[NEW] == R_PRIMARY;
struct drbd_peer_device *peer_device;
for_each_peer_device(peer_device, device) {
- if (peer_device->connection->peer_role[NOW] == R_PRIMARY)
+ struct drbd_connection *connection = peer_device->connection;
+
+ /* Do not consider the peer if we are disconnecting. */
+ if (resource->remote_state_change &&
+ drbd_twopc_between_peer_and_me(connection) &&
+ resource->twopc_reply.is_disconnect)
+ continue;
+
+ if (connection->peer_role[NEW] == R_PRIMARY)
primary = true;
- if (peer_device->disk_state[NOW] == D_UP_TO_DATE)
+ if (peer_device->disk_state[NEW] == D_UP_TO_DATE)
up_to_date_data = true;
}
@@ -4808,6 +4817,7 @@ change_cluster_wide_state(bool (*change)(struct change_context *, enum change_ph
} else if (context->mask.conn == conn_MASK && context->val.conn == C_DISCONNECTING) {
reply->target_reachable_nodes = NODE_MASK(context->target_node_id);
reply->reachable_nodes &= ~reply->target_reachable_nodes;
+ reply->is_disconnect = 1;
} else {
reply->target_reachable_nodes = reply->reachable_nodes;
}
--
2.35.3

32
0003-drbd-fix-error-path-in-drbd_get_listener.patch Normal file
View File

@ -0,0 +1,32 @@
From fbfb92d11e64daec167b24521c715ceab505b55d Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Thu, 7 Sep 2023 10:36:29 +0200
Subject: [PATCH 03/20] drbd: fix error path in drbd_get_listener()
When initializing a listener fails do not access the fields of the
listener struct after giving up the reference.
---
drbd/drbd_transport.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drbd/drbd_transport.c b/drbd/drbd_transport.c
index 416a50499046..2aefd71ee395 100644
--- a/drbd/drbd_transport.c
+++ b/drbd/drbd_transport.c
@@ -207,11 +207,11 @@ int drbd_get_listener(struct drbd_transport *transport, struct drbd_path *path,
if (needs_init) {
err = init_listener(transport, addr, &init_net, listener);
+ listener->err = err;
+ complete_all(&listener->ready);
if (err)
drbd_put_listener(path);
- listener->err = err;
- complete_all(&listener->ready);
return err;
}
--
2.35.3

47
0004-drbd-build-fix-spurious-re-build-attempt-of-compat.p.patch Normal file
View File

@ -0,0 +1,47 @@
From 67ac093d6dc176fd6a3bf0c7f5a3ad046d48f558 Mon Sep 17 00:00:00 2001
From: Lars Ellenberg <lars.ellenberg@linbit.com>
Date: Wed, 6 Sep 2023 14:38:32 +0200
Subject: [PATCH 04/20] drbd: build: fix spurious re-build attempt of
compat.patch
Patching changed the timestamp of the patched files,
which are pre-requisites for generating the patch.
The second build after extracting the tarball
would try to regenerate the patch file.
Exclude generated *.mod.c from "to-be-patched" files for spatch.
Reset timestamp of patched files to that of the patch file, instead of current time.
---
drbd/Kbuild | 1 +
drbd/Makefile | 2 +-
2 files changed, 2 insertions(+), 1 deletion(-)
diff --git a/drbd/Kbuild b/drbd/Kbuild
index 1576dc19e1da..073469ae4191 100644
--- a/drbd/Kbuild
+++ b/drbd/Kbuild
@@ -110,6 +110,7 @@ filechk_compat.h = cat $(obj)/compat.$(KERNELVERSION).h
$(src)/.compat_patches_applied: $(src)/drbd-kernel-compat/compat.patch
@$(kecho) ' PATCH'
@patch -d $(src) -p0 --batch --forward --reject-file=- < $(src)/drbd-kernel-compat/compat.patch
+ @cd $(src) && p=drbd-kernel-compat/compat.patch && sed -ne 's/^--- //p' < $$p | xargs -r -- touch -r $$p
@cp -d $(src)/drbd-kernel-compat/compat.patch $(src)/.compat_patches_applied
$(src)/drbd-kernel-compat/compat.patch: $(obj)/compat.h
diff --git a/drbd/Makefile b/drbd/Makefile
index 09a1efbf7fe1..ecdff04b9808 100644
--- a/drbd/Makefile
+++ b/drbd/Makefile
@@ -170,7 +170,7 @@ else
compat_headers := $(wildcard drbd-kernel-compat/cocci_cache/*/compat.h)
compat_patches := $(patsubst %.h,%.patch,$(compat_headers))
- sources := $(filter-out drbd_strings.c drbd_buildtag.c,$(wildcard *.c))
+ sources := $(filter-out drbd_strings.c drbd_buildtag.c drbd%.mod.c,$(wildcard *.c))
sources += $(wildcard drbd-headers/linux/*.h)
$(compat_patches): $(sources)
--
2.35.3

28
0005-drbd-log-error-code-when-thread-fails-to-start.patch Normal file
View File

@ -0,0 +1,28 @@
From a98c818bd33920fd5189cdd8f5d81850ad4a945b Mon Sep 17 00:00:00 2001
From: Joel Colledge <joel.colledge@linbit.com>
Date: Thu, 7 Sep 2023 17:32:56 +0200
Subject: [PATCH 05/20] drbd: log error code when thread fails to start
---
drbd/drbd_main.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index 38f728f3da73..810df864f60b 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -642,9 +642,9 @@ int drbd_thread_start(struct drbd_thread *thi)
if (IS_ERR(nt)) {
if (connection)
- drbd_err(connection, "Couldn't start thread\n");
+ drbd_err(connection, "Couldn't start thread: %ld\n", PTR_ERR(nt));
else
- drbd_err(resource, "Couldn't start thread\n");
+ drbd_err(resource, "Couldn't start thread: %ld\n", PTR_ERR(nt));
return false;
}
--
2.35.3

82
0006-drbd-log-numeric-value-of-drbd_state_rv-as-well-as-s.patch Normal file
View File

@ -0,0 +1,82 @@
From 8c2c783a09d81f0a725e7a3ae48be4ceb2d79a63 Mon Sep 17 00:00:00 2001
From: Joel Colledge <joel.colledge@linbit.com>
Date: Fri, 8 Sep 2023 11:26:01 +0200
Subject: [PATCH 06/20] drbd: log numeric value of drbd_state_rv as well as
string form
"Auto-promote failed: ?" was seen in a log. Logging the numeric value of
the state change return value gives us more information about what
happened in such a case.
---
drbd/drbd_main.c | 8 ++++----
drbd/drbd_receiver.c | 6 +++---
drbd/drbd_state.c | 3 ++-
3 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index 810df864f60b..e26cf7e2b008 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -2698,8 +2698,8 @@ static int drbd_open(struct block_device *bdev, fmode_t mode)
if (resource->role[NOW] == R_SECONDARY) {
rv = try_to_promote(device, timeout, (mode & FMODE_NDELAY));
if (rv < SS_SUCCESS)
- drbd_info(resource, "Auto-promote failed: %s\n",
- drbd_set_st_err_str(rv));
+ drbd_info(resource, "Auto-promote failed: %s (%d)\n",
+ drbd_set_st_err_str(rv), rv);
}
} else if ((mode & FMODE_NDELAY) == 0) {
/* Double check peers
@@ -2856,8 +2856,8 @@ static void drbd_release(struct gendisk *gd, fmode_t mode)
!test_bit(EXPLICIT_PRIMARY, &resource->flags)) {
rv = drbd_set_role(resource, R_SECONDARY, false, NULL);
if (rv < SS_SUCCESS)
- drbd_warn(resource, "Auto-demote failed: %s\n",
- drbd_set_st_err_str(rv));
+ drbd_warn(resource, "Auto-demote failed: %s (%d)\n",
+ drbd_set_st_err_str(rv), rv);
}
}
diff --git a/drbd/drbd_receiver.c b/drbd/drbd_receiver.c
index 95cf7ac36762..2162d13cb25e 100644
--- a/drbd/drbd_receiver.c
+++ b/drbd/drbd_receiver.c
@@ -983,8 +983,8 @@ static int connect_work(struct drbd_work *work, int cancel)
drbd_send_disconnect(connection);
apply_local_state_change(connection, OUTDATE_DISKS_AND_DISCONNECT, force_demote);
} else {
- drbd_info(connection, "Failure to connect %d %s; retrying\n",
- rv, drbd_set_st_err_str(rv));
+ drbd_info(connection, "Failure to connect: %s (%d); retrying\n",
+ drbd_set_st_err_str(rv), rv);
change_cstate(connection, C_NETWORK_FAILURE, CS_HARD);
}
@@ -6107,7 +6107,7 @@ out:
}
if (rv < SS_SUCCESS) {
- drbd_err(resource, "State change failed: %s\n", drbd_set_st_err_str(rv));
+ drbd_err(resource, "State change failed: %s (%d)\n", drbd_set_st_err_str(rv), rv);
if (rv == SS_PRIMARY_READER)
log_openers(resource);
}
diff --git a/drbd/drbd_state.c b/drbd/drbd_state.c
index 8b60afeb097b..23eab7f867aa 100644
--- a/drbd/drbd_state.c
+++ b/drbd/drbd_state.c
@@ -791,7 +791,8 @@ static enum drbd_state_rv ___end_state_change(struct drbd_resource *resource, st
rv = try_state_change(resource);
if (rv < SS_SUCCESS) {
if (flags & CS_VERBOSE) {
- drbd_err(resource, "State change failed: %s\n", drbd_set_st_err_str(rv));
+ drbd_err(resource, "State change failed: %s (%d)\n",
+ drbd_set_st_err_str(rv), rv);
print_state_change(resource, "Failed: ");
}
goto out;
--
2.35.3

45
0007-drbd-stop-defining-__KERNEL_SYSCALLS__.patch Normal file
View File

@ -0,0 +1,45 @@
From 15cf257c277020e1ba76eff7aa99ed08d44f7d3c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Moritz=20=22WanzenBug=22=20Wanzenb=C3=B6ck?=
<moritz.wanzenboeck@linbit.com>
Date: Fri, 8 Sep 2023 11:45:48 +0200
Subject: [PATCH 07/20] drbd: stop defining __KERNEL_SYSCALLS__
See also upstream Linux kernel commit
d519df00938e ("drbd: stop defining __KERNEL_SYSCALLS__")
Original message:
__KERNEL_SYSCALLS__ hasn't been needed since Linux 2.6.19 so stop
defining it.
---
drbd/drbd_main.c | 1 -
drbd/drbd_receiver.c | 1 -
2 files changed, 2 deletions(-)
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index e26cf7e2b008..ccf0f72829db 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -36,7 +36,6 @@
#include <linux/notifier.h>
#include <linux/workqueue.h>
#include <linux/kthread.h>
-#define __KERNEL_SYSCALLS__
#include <linux/unistd.h>
#include <linux/vmalloc.h>
#include <linux/device.h>
diff --git a/drbd/drbd_receiver.c b/drbd/drbd_receiver.c
index 2162d13cb25e..2e2ed4699a94 100644
--- a/drbd/drbd_receiver.c
+++ b/drbd/drbd_receiver.c
@@ -26,7 +26,6 @@
#include <linux/slab.h>
#include <linux/pkt_sched.h>
#include <uapi/linux/sched/types.h>
-#define __KERNEL_SYSCALLS__
#include <linux/unistd.h>
#include <linux/vmalloc.h>
#include <linux/random.h>
--
2.35.3

80
0008-compat-block-introduce-holder-ops.patch Normal file
View File

@ -0,0 +1,80 @@
From c34a13c3df85352124e94456f81e3d4ba4f52440 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Moritz=20=22WanzenBug=22=20Wanzenb=C3=B6ck?=
<moritz.wanzenboeck@linbit.com>
Date: Fri, 8 Sep 2023 12:20:10 +0200
Subject: [PATCH 08/20] compat: block: introduce holder ops
See also upstream Linux kernel commit
0718afd47f70 ("block: introduce holder ops")
Original message:
Add a new blk_holder_ops structure, which is passed to blkdev_get_by_* and
installed in the block_device for exclusive claims. It will be used to
allow the block layer to call back into the user of the block device for
thing like notification of a removed device or a device resize.
---
.../cocci/blkdev_get_by_path__no_has_holder_ops.cocci | 6 ++++++
drbd/drbd-kernel-compat/gen_patch_names.c | 3 +++
.../tests/blkdev_get_by_path_has_holder_ops.c | 7 +++++++
drbd/drbd_nl.c | 3 ++-
4 files changed, 18 insertions(+), 1 deletion(-)
create mode 100644 drbd/drbd-kernel-compat/cocci/blkdev_get_by_path__no_has_holder_ops.cocci
create mode 100644 drbd/drbd-kernel-compat/tests/blkdev_get_by_path_has_holder_ops.c
diff --git a/drbd/drbd-kernel-compat/cocci/blkdev_get_by_path__no_has_holder_ops.cocci b/drbd/drbd-kernel-compat/cocci/blkdev_get_by_path__no_has_holder_ops.cocci
new file mode 100644
index 000000000000..050d89e6fe0d
--- /dev/null
+++ b/drbd/drbd-kernel-compat/cocci/blkdev_get_by_path__no_has_holder_ops.cocci
@@ -0,0 +1,6 @@
+@@
+expression path, mode, holder, ops;
+@@
+ blkdev_get_by_path(path, mode, holder
+- , ops
+ )
diff --git a/drbd/drbd-kernel-compat/gen_patch_names.c b/drbd/drbd-kernel-compat/gen_patch_names.c
index 959bbc351e1b..7071a0a4c5ec 100644
--- a/drbd/drbd-kernel-compat/gen_patch_names.c
+++ b/drbd/drbd-kernel-compat/gen_patch_names.c
@@ -556,6 +556,9 @@ int main(int argc, char **argv)
patch(1, "__bio_add_page", true, false,
COMPAT_HAVE___BIO_ADD_PAGE, "present");
+ patch(1, "blkdev_get_by_path", true, false,
+ COMPAT_BLKDEV_GET_BY_PATH_HAS_HOLDER_OPS, "has_holder_ops");
+
/* #define BLKDEV_ISSUE_ZEROOUT_EXPORTED */
/* #define BLKDEV_ZERO_NOUNMAP */
diff --git a/drbd/drbd-kernel-compat/tests/blkdev_get_by_path_has_holder_ops.c b/drbd/drbd-kernel-compat/tests/blkdev_get_by_path_has_holder_ops.c
new file mode 100644
index 000000000000..02a560782f37
--- /dev/null
+++ b/drbd/drbd-kernel-compat/tests/blkdev_get_by_path_has_holder_ops.c
@@ -0,0 +1,7 @@
+/* { "version": "v6.5-rc1", "commit": "0718afd47f70cf46877c39c25d06b786e1a3f36c", "comment": "block: introduce holder ops", "author": "Christoph Hellwig <hch@lst.de>", "date": "Thu Jun 1 11:44:52 2023 +0200" } */
+#include <linux/blkdev.h>
+
+struct block_device *foo(const char *bdev_path, struct blk_holder_ops *ops)
+{
+ return blkdev_get_by_path(bdev_path, 0, NULL, ops);
+}
diff --git a/drbd/drbd_nl.c b/drbd/drbd_nl.c
index 0fc1d84a996f..0a67bfa4ca52 100644
--- a/drbd/drbd_nl.c
+++ b/drbd/drbd_nl.c
@@ -2560,7 +2560,8 @@ static struct block_device *open_backing_dev(struct drbd_device *device,
const char *bdev_path, void *claim_ptr)
{
struct block_device *bdev = blkdev_get_by_path(bdev_path,
- FMODE_READ | FMODE_WRITE | FMODE_EXCL, claim_ptr);
+ FMODE_READ | FMODE_WRITE | FMODE_EXCL,
+ claim_ptr, NULL);
if (IS_ERR(bdev)) {
drbd_err(device, "open(\"%s\") failed with %ld\n",
bdev_path, PTR_ERR(bdev));
--
2.35.3

45
0009-drbd-reduce-net_ee-not-empty-info-to-a-dynamic-debug.patch Normal file
View File

@ -0,0 +1,45 @@
From fc7b1ad2c422e4148d6419f1eec56189a7a578ac Mon Sep 17 00:00:00 2001
From: Joel Colledge <joel.colledge@linbit.com>
Date: Fri, 1 Sep 2023 11:04:49 +0200
Subject: [PATCH 09/20] drbd: reduce "net_ee not empty" info to a dynamic debug
print
This situation is known and harmless. No need to print a worrying
"killed N entries" message when it occurs.
A similar comment explaining that this is harmless was introduced in
commit:
31e0f1250f17 on disconnect, just give up all references on pages from net_ee
But the comment got lost in some refactoring:
0b220db6109e drbd: move {active,read,sync,done}_ee lists from device to connection
Bring the comment back, slightly simplified.
---
drbd/drbd_receiver.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/drbd/drbd_receiver.c b/drbd/drbd_receiver.c
index 2e2ed4699a94..223353b1081c 100644
--- a/drbd/drbd_receiver.c
+++ b/drbd/drbd_receiver.c
@@ -8664,9 +8664,15 @@ static void conn_disconnect(struct drbd_connection *connection)
i = drbd_free_peer_reqs(connection, &connection->sync_ee);
if (i)
drbd_info(connection, "sync_ee not empty, killed %u entries\n", i);
+
+ /*
+ * tcp_close and release of sendpage pages can be deferred. We don't
+ * care for exactly when the network stack does its put_page(), but
+ * release our reference on these pages right here.
+ */
i = drbd_free_peer_reqs(connection, &connection->net_ee);
if (i)
- drbd_info(connection, "net_ee not empty, killed %u entries\n", i);
+ dynamic_drbd_dbg(connection, "net_ee not empty, killed %u entries\n", i);
cleanup_unacked_peer_requests(connection);
cleanup_peer_ack_list(connection);
--
2.35.3

37
0010-drbd-do-not-send-P_CURRENT_UUID-to-DRBD-8-peer-when-.patch Normal file
View File

@ -0,0 +1,37 @@
From 1b94c10250c7d37423ba894457c40d9799025a93 Mon Sep 17 00:00:00 2001
From: Joel Colledge <joel.colledge@linbit.com>
Date: Mon, 11 Sep 2023 09:20:24 +0200
Subject: [PATCH 10/20] drbd: do not send P_CURRENT_UUID to DRBD 8 peer when
our disk fails
DRBD 8 does not understand P_CURRENT_UUID. Skip the peer in
diskfull_peers_need_new_cur_uuid() to avoid sending such a packet.
This is valid because the DRBD 8 peer generates a new UUID itself when
our disk fails.
This prevents failures of the form:
drbd res: Unexpected data packet Unknown (0x0044)
drbd res: peer( Primary -> Unknown ) conn( Connected -> ProtocolError ) pdsk( Failed -> DUnknown )
---
drbd/drbd_main.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index ccf0f72829db..e6ed1185c710 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -4614,6 +4614,9 @@ static bool diskfull_peers_need_new_cur_uuid(struct drbd_device *device)
rcu_read_lock();
for_each_peer_device_rcu(peer_device, device) {
+ if (peer_device->connection->agreed_pro_version < 110)
+ continue;
+
/* Only an up-to-date peer persists a new current uuid! */
if (peer_device->disk_state[NOW] < D_UP_TO_DATE)
continue;
--
2.35.3

129
0011-compat-block-pass-a-gendisk-to-open.patch Normal file
View File

@ -0,0 +1,129 @@
From b993152e80676f5b9ce583c9471b630cbd0675d7 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Moritz=20=22WanzenBug=22=20Wanzenb=C3=B6ck?=
<moritz.wanzenboeck@linbit.com>
Date: Fri, 8 Sep 2023 15:09:47 +0200
Subject: [PATCH 11/20] compat: block: pass a gendisk to ->open
See also upstream Linux kernel commit
d32e2bf83791 ("block: pass a gendisk to ->open")
Original message:
->open is only called on the whole device. Make that explicit by
passing a gendisk instead of the block_device.
---
...ice_operations_open__no_takes_gendisk.cocci | 18 ++++++++++++++++++
drbd/drbd-kernel-compat/gen_patch_names.c | 3 +++
...lock_device_operations_open_takes_gendisk.c | 17 +++++++++++++++++
drbd/drbd_main.c | 10 +++++-----
4 files changed, 43 insertions(+), 5 deletions(-)
create mode 100644 drbd/drbd-kernel-compat/cocci/block_device_operations_open__no_takes_gendisk.cocci
create mode 100644 drbd/drbd-kernel-compat/tests/block_device_operations_open_takes_gendisk.c
diff --git a/drbd/drbd-kernel-compat/cocci/block_device_operations_open__no_takes_gendisk.cocci b/drbd/drbd-kernel-compat/cocci/block_device_operations_open__no_takes_gendisk.cocci
new file mode 100644
index 000000000000..2ba2856da9db
--- /dev/null
+++ b/drbd/drbd-kernel-compat/cocci/block_device_operations_open__no_takes_gendisk.cocci
@@ -0,0 +1,18 @@
+@ drbd_open_arg @
+identifier gd;
+fresh identifier bdev = "" ## "bdev";
+@@
+ drbd_open(
+- struct gendisk *gd,
++ struct block_device *bdev,
+ ... ) {
+<...
+(
+- gd->part0
++ bdev
+|
+- gd
++ bdev->bd_disk
+)
+...>
+}
diff --git a/drbd/drbd-kernel-compat/gen_patch_names.c b/drbd/drbd-kernel-compat/gen_patch_names.c
index 7071a0a4c5ec..6e4f06d9a3a7 100644
--- a/drbd/drbd-kernel-compat/gen_patch_names.c
+++ b/drbd/drbd-kernel-compat/gen_patch_names.c
@@ -559,6 +559,9 @@ int main(int argc, char **argv)
patch(1, "blkdev_get_by_path", true, false,
COMPAT_BLKDEV_GET_BY_PATH_HAS_HOLDER_OPS, "has_holder_ops");
+ patch(1, "block_device_operations_open", true, false,
+ COMPAT_BLOCK_DEVICE_OPERATIONS_OPEN_TAKES_GENDISK, "takes_gendisk");
+
/* #define BLKDEV_ISSUE_ZEROOUT_EXPORTED */
/* #define BLKDEV_ZERO_NOUNMAP */
diff --git a/drbd/drbd-kernel-compat/tests/block_device_operations_open_takes_gendisk.c b/drbd/drbd-kernel-compat/tests/block_device_operations_open_takes_gendisk.c
new file mode 100644
index 000000000000..d5f20fd569fb
--- /dev/null
+++ b/drbd/drbd-kernel-compat/tests/block_device_operations_open_takes_gendisk.c
@@ -0,0 +1,17 @@
+/* { "version": "v6.5-rc1", "commit": "d32e2bf83791727a84ad5d3e3d713e82f9adbe30", "comment": "block: pass a gendisk to ->open", "author": "Christoph Hellwig <hch@lst.de>", "date": "Thu Jun 8 13:02:36 2023 +0200" } */
+#include <linux/blkdev.h>
+
+#ifndef __same_type
+# define __same_type(a, b) __builtin_types_compatible_p(typeof(a), typeof(b))
+#endif
+
+int foo_open(struct gendisk *disk, unsigned int mode)
+{
+ return 0;
+}
+
+void foo(void)
+{
+ struct block_device_operations ops;
+ BUILD_BUG_ON(!(__same_type(ops.open, &foo_open)));
+}
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index e6ed1185c710..53cb4c9e7db3 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -52,7 +52,7 @@
#include "drbd_meta_data.h"
#include "drbd_dax_pmem.h"
-static int drbd_open(struct block_device *bdev, fmode_t mode);
+static int drbd_open(struct gendisk *gd, fmode_t mode);
static void drbd_release(struct gendisk *gd, fmode_t mode);
static void md_sync_timer_fn(struct timer_list *t);
static int w_bitmap_io(struct drbd_work *w, int unused);
@@ -2646,9 +2646,9 @@ out:
spin_unlock(&device->openers_lock);
}
-static int drbd_open(struct block_device *bdev, fmode_t mode)
+static int drbd_open(struct gendisk *gd, fmode_t mode)
{
- struct drbd_device *device = bdev->bd_disk->private_data;
+ struct drbd_device *device = gd->private_data;
struct drbd_resource *resource = device->resource;
long timeout = resource->res_opts.auto_promote_timeout * HZ / 10;
enum ioc_rv r;
@@ -2664,7 +2664,7 @@ static int drbd_open(struct block_device *bdev, fmode_t mode)
/* Fail read-write open early,
* in case someone explicitly set us read-only (blockdev --setro) */
- if (bdev_read_only(bdev) && (mode & FMODE_WRITE))
+ if (bdev_read_only(gd->part0) && (mode & FMODE_WRITE))
return -EACCES;
if (resource->fail_io[NOW])
@@ -2740,7 +2740,7 @@ out:
mutex_unlock(&resource->open_release);
if (err) {
- drbd_release(bdev->bd_disk, mode);
+ drbd_release(gd, mode);
if (err == -EAGAIN && !(mode & FMODE_NDELAY))
err = -EMEDIUMTYPE;
}
--
2.35.3

67
0012-drbd-Restore-DATA_CORKED-and-CONTROL_CORKED-bits.patch Normal file
View File

@ -0,0 +1,67 @@
From 88bf7f95dc359f4901f08ac8fb8ffc7dfceca9c1 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Wed, 13 Sep 2023 17:50:37 +0200
Subject: [PATCH 12/20] drbd: Restore DATA_CORKED and CONTROL_CORKED bits
They were removed with 9072e1d1b6b3 in January 2022.
Since then, the CONTROL_CORKED mapped onto C_UNREGISTERED.
I think that had a negligible impact because C_UNREGISTERED indicates
if a connection was deleted before. It is no longer connected when it
is time to delete a connection. Therefore, the corking code can not
set that double-used bit.
The other user of C_UNREGISTERED is when iterating over connections. A
wrongly set C_UNREGISTERED causes another iteration over the
connections list - no further harm.
Still, I am fixing this by allocating dedicated bits for corking in
the flags word.
---
drbd/drbd_debugfs.c | 2 ++
drbd/drbd_int.h | 2 ++
drbd/drbd_main.c | 1 +
3 files changed, 5 insertions(+)
diff --git a/drbd/drbd_debugfs.c b/drbd/drbd_debugfs.c
index 778a995018d7..b8e4f2cde426 100644
--- a/drbd/drbd_debugfs.c
+++ b/drbd/drbd_debugfs.c
@@ -839,6 +839,8 @@ static int connection_debug_show(struct seq_file *m, void *ignored)
pretty_print_bit(CONN_DRY_RUN);
pretty_print_bit(DISCONNECT_EXPECTED);
pretty_print_bit(BARRIER_ACK_PENDING);
+ pretty_print_bit(DATA_CORKED);
+ pretty_print_bit(CONTROL_CORKED);
pretty_print_bit(C_UNREGISTERED);
pretty_print_bit(RECONNECT);
pretty_print_bit(CONN_DISCARD_MY_DATA);
diff --git a/drbd/drbd_int.h b/drbd/drbd_int.h
index 17be90e7d8fa..4f3228d5a4d1 100644
--- a/drbd/drbd_int.h
+++ b/drbd/drbd_int.h
@@ -755,6 +755,8 @@ enum connection_flag {
DISCONNECT_EXPECTED,
BARRIER_ACK_PENDING,
CORKED,
+ DATA_CORKED = CORKED, /* used as computed value CORKED + DATA_STREAM */
+ CONTROL_CORKED, /* used as computed value CORKED + CONTROL_STREAM */
C_UNREGISTERED,
RECONNECT,
CONN_DISCARD_MY_DATA,
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index 53cb4c9e7db3..6bb618909aa0 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -1039,6 +1039,7 @@ int __send_command(struct drbd_connection *connection, int vnr,
struct drbd_send_buffer *sbuf = &connection->send_buffer[drbd_stream];
struct drbd_transport *transport = &connection->transport;
struct drbd_transport_ops *tr_ops = transport->ops;
+ /* CORKED + drbd_stream is either DATA_CORKED or CONTROL_CORKED */
bool corked = test_bit(CORKED + drbd_stream, &connection->flags);
bool flush = stream_and_flags & SFLAG_FLUSH;
int err;
--
2.35.3

38
0013-drbd-remove-unused-extern-for-conn_try_outdate_peer.patch Normal file
View File

@ -0,0 +1,38 @@
From 9e087399533445e3df60a34f26a1e6a285b0504a Mon Sep 17 00:00:00 2001
From: Joel Colledge <joel.colledge@linbit.com>
Date: Tue, 12 Sep 2023 16:36:43 +0200
Subject: [PATCH 13/20] drbd: remove unused extern for conn_try_outdate_peer()
---
drbd/drbd_int.h | 1 -
drbd/drbd_nl.c | 2 +-
2 files changed, 1 insertion(+), 2 deletions(-)
diff --git a/drbd/drbd_int.h b/drbd/drbd_int.h
index 4f3228d5a4d1..f6e7c3ac2629 100644
--- a/drbd/drbd_int.h
+++ b/drbd/drbd_int.h
@@ -1932,7 +1932,6 @@ extern void drbd_reconsider_queue_parameters(struct drbd_device *device,
struct drbd_backing_dev *bdev);
extern bool barrier_pending(struct drbd_resource *resource);
extern enum drbd_state_rv drbd_set_role(struct drbd_resource *, enum drbd_role, bool, struct sk_buff *);
-extern bool conn_try_outdate_peer(struct drbd_connection *connection);
extern void conn_try_outdate_peer_async(struct drbd_connection *connection);
extern int drbd_maybe_khelper(struct drbd_device *, struct drbd_connection *, char *);
extern int drbd_create_peer_device_default_config(struct drbd_peer_device *peer_device);
diff --git a/drbd/drbd_nl.c b/drbd/drbd_nl.c
index 0a67bfa4ca52..cb5cdb184824 100644
--- a/drbd/drbd_nl.c
+++ b/drbd/drbd_nl.c
@@ -768,7 +768,7 @@ static bool intentional_diskless(struct drbd_resource *resource)
return intentional_diskless;
}
-bool conn_try_outdate_peer(struct drbd_connection *connection)
+static bool conn_try_outdate_peer(struct drbd_connection *connection)
{
struct drbd_resource *resource = connection->resource;
unsigned long last_reconnect_jif;
--
2.35.3

1480
0014-drbd-include-source-of-state-change-in-log.patch Normal file
View File

File diff suppressed because it is too large Load Diff

202
0015-compat-block-use-the-holder-as-indication-for-exclus.patch Normal file
View File

@ -0,0 +1,202 @@
From 3b9fcc2cfaa32766724f371cc2054e057adbc425 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Moritz=20=22WanzenBug=22=20Wanzenb=C3=B6ck?=
<moritz.wanzenboeck@linbit.com>
Date: Mon, 11 Sep 2023 13:36:07 +0200
Subject: [PATCH 15/20] compat: block: use the holder as indication for
exclusive opens
See also upstream Linux kernel commit
2736e8eeb0cc ("block: use the holder as indication for exclusive opens")
Original message:
The current interface for exclusive opens is rather confusing as it
requires both the FMODE_EXCL flag and a holder. Remove the need to pass
FMODE_EXCL and just key off the exclusive open off a non-NULL holder.
For blkdev_put this requires adding the holder argument, which provides
better debug checking that only the holder actually releases the hold,
but at the same time allows removing the now superfluous mode argument.
---
.../cocci/blkdev_put__no_has_holder.cocci | 38 +++++++++++++++++++
drbd/drbd-kernel-compat/gen_patch_names.c | 3 ++
.../tests/blkdev_put_has_holder.c | 17 +++++++++
drbd/drbd_nl.c | 28 ++++++++------
4 files changed, 75 insertions(+), 11 deletions(-)
create mode 100644 drbd/drbd-kernel-compat/cocci/blkdev_put__no_has_holder.cocci
create mode 100644 drbd/drbd-kernel-compat/tests/blkdev_put_has_holder.c
diff --git a/drbd/drbd-kernel-compat/cocci/blkdev_put__no_has_holder.cocci b/drbd/drbd-kernel-compat/cocci/blkdev_put__no_has_holder.cocci
new file mode 100644
index 000000000000..c903bc2d529c
--- /dev/null
+++ b/drbd/drbd-kernel-compat/cocci/blkdev_put__no_has_holder.cocci
@@ -0,0 +1,38 @@
+@@
+expression path, mode;
+@@
+ blkdev_get_by_path(
+ path,
+- mode,
++ mode | FMODE_EXCL,
+ ...
+ )
+
+@@
+expression bdev, holder;
+@@
+ blkdev_put(
+ bdev,
+- holder
++ FMODE_READ | FMODE_WRITE | FMODE_EXCL
+ )
+
+@@
+identifier device, bdev, holder, do_bd_unlink;
+@@
+ void close_backing_dev(
+ struct drbd_device *device,
+ struct block_device *bdev,
+- void *holder,
+ bool do_bd_unlink
+ ) { ... }
+
+@@
+expression device, bdev, holder, do_bd_unlink;
+@@
+ close_backing_dev(
+ device,
+ bdev,
+- holder,
+ do_bd_unlink
+ )
diff --git a/drbd/drbd-kernel-compat/gen_patch_names.c b/drbd/drbd-kernel-compat/gen_patch_names.c
index 6e4f06d9a3a7..4761c1ef7d0c 100644
--- a/drbd/drbd-kernel-compat/gen_patch_names.c
+++ b/drbd/drbd-kernel-compat/gen_patch_names.c
@@ -562,6 +562,9 @@ int main(int argc, char **argv)
patch(1, "block_device_operations_open", true, false,
COMPAT_BLOCK_DEVICE_OPERATIONS_OPEN_TAKES_GENDISK, "takes_gendisk");
+ patch(1, "blkdev_put", true, false,
+ COMPAT_BLKDEV_PUT_HAS_HOLDER, "has_holder");
+
/* #define BLKDEV_ISSUE_ZEROOUT_EXPORTED */
/* #define BLKDEV_ZERO_NOUNMAP */
diff --git a/drbd/drbd-kernel-compat/tests/blkdev_put_has_holder.c b/drbd/drbd-kernel-compat/tests/blkdev_put_has_holder.c
new file mode 100644
index 000000000000..d5f0c5dd0355
--- /dev/null
+++ b/drbd/drbd-kernel-compat/tests/blkdev_put_has_holder.c
@@ -0,0 +1,17 @@
+/* { "version": "v6.5-rc1", "commit": "ae220766d87cd6799dbf918fea10613ae14c0654", "comment": "block: remove the unused mode argument to ->release", "author": "Christoph Hellwig <hch@lst.de>", "date": "Thu Jun 8 13:02:37 2023 +0200" } */
+#include <linux/blkdev.h>
+
+#ifndef __same_type
+# define __same_type(a, b) __builtin_types_compatible_p(typeof(a), typeof(b))
+#endif
+
+void foo_blkdev_put(struct block_device *bdev, void *holder)
+{
+}
+
+
+void foo(void)
+{
+ BUILD_BUG_ON(!(__same_type(&blkdev_put, &foo_blkdev_put)));
+}
+
diff --git a/drbd/drbd_nl.c b/drbd/drbd_nl.c
index b7e9e43312f9..8c968cf252ca 100644
--- a/drbd/drbd_nl.c
+++ b/drbd/drbd_nl.c
@@ -2536,13 +2536,13 @@ bool want_bitmap(struct drbd_peer_device *peer_device)
}
static void close_backing_dev(struct drbd_device *device, struct block_device *bdev,
- bool do_bd_unlink)
+ void *holder, bool do_bd_unlink)
{
if (!bdev)
return;
if (do_bd_unlink)
bd_unlink_disk_holder(bdev, device->vdisk);
- blkdev_put(bdev, FMODE_READ | FMODE_WRITE | FMODE_EXCL);
+ blkdev_put(bdev, holder);
}
void drbd_backing_dev_free(struct drbd_device *device, struct drbd_backing_dev *ldev)
@@ -2552,8 +2552,11 @@ void drbd_backing_dev_free(struct drbd_device *device, struct drbd_backing_dev *
drbd_dax_close(ldev);
- close_backing_dev(device, ldev->md_bdev, ldev->md_bdev != ldev->backing_bdev);
- close_backing_dev(device, ldev->backing_bdev, true);
+ close_backing_dev(device,
+ ldev->md_bdev,
+ ldev->md.meta_dev_idx < 0 ? (void *)device : (void *)drbd_m_holder,
+ ldev->md_bdev != ldev->backing_bdev);
+ close_backing_dev(device, ldev->backing_bdev, device, true);
kfree(ldev->disk_conf);
kfree(ldev);
@@ -2563,7 +2566,7 @@ static struct block_device *open_backing_dev(struct drbd_device *device,
const char *bdev_path, void *claim_ptr)
{
struct block_device *bdev = blkdev_get_by_path(bdev_path,
- FMODE_READ | FMODE_WRITE | FMODE_EXCL,
+ FMODE_READ | FMODE_WRITE,
claim_ptr, NULL);
if (IS_ERR(bdev)) {
drbd_err(device, "open(\"%s\") failed with %ld\n",
@@ -2588,6 +2591,7 @@ static int open_backing_devices(struct drbd_device *device,
struct drbd_backing_dev *nbc)
{
struct block_device *bdev;
+ void *meta_claim_ptr;
int err;
bdev = open_backing_dev(device, new_disk_conf->backing_dev, device);
@@ -2597,12 +2601,17 @@ static int open_backing_devices(struct drbd_device *device,
err = link_backing_dev(device, new_disk_conf->backing_dev, bdev);
if (err) {
/* close without unlinking; otherwise error path will try to unlink */
- close_backing_dev(device, bdev, false);
+ close_backing_dev(device, bdev, device, false);
return ERR_OPEN_DISK;
}
nbc->backing_bdev = bdev;
+ /* meta_claim_ptr: device, if claimed exclusively; shared drbd_m_holder,
+ * if potentially shared with other drbd minors
+ */
+ meta_claim_ptr = (new_disk_conf->meta_dev_idx < 0) ?
+ (void *)device : (void *)drbd_m_holder;
/*
* meta_dev_idx >= 0: external fixed size, possibly multiple
* drbd sharing one meta device. TODO in that case, paranoia
@@ -2611,10 +2620,7 @@ static int open_backing_devices(struct drbd_device *device,
* should check it for you already; but if you don't, or
* someone fooled it, we need to double check here)
*/
- bdev = open_backing_dev(device, new_disk_conf->meta_dev,
- /* claim ptr: device, if claimed exclusively; shared drbd_m_holder,
- * if potentially shared with other drbd minors */
- (new_disk_conf->meta_dev_idx < 0) ? (void*)device : (void*)drbd_m_holder);
+ bdev = open_backing_dev(device, new_disk_conf->meta_dev, meta_claim_ptr);
if (IS_ERR(bdev))
return ERR_OPEN_MD_DISK;
@@ -2624,7 +2630,7 @@ static int open_backing_devices(struct drbd_device *device,
err = link_backing_dev(device, new_disk_conf->meta_dev, bdev);
if (err) {
/* close without unlinking; otherwise error path will try to unlink */
- close_backing_dev(device, bdev, false);
+ close_backing_dev(device, bdev, meta_claim_ptr, false);
return ERR_OPEN_MD_DISK;
}
}
--
2.35.3

33
0016-drbd-Fix-net-options-set-defaults-to-not-clear-the-t.patch Normal file
View File

@ -0,0 +1,33 @@
From 69d9286628b730870665cd2c6f05dd9f1813c65e Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Sun, 17 Sep 2023 22:21:11 +0800
Subject: [PATCH 16/20] drbd: Fix `net-options --set-defaults` to not clear the
transport
So far, `drbdsetup net-options --set-defaults` cleared the
transport_name. That is a stupid bug that caused `drbdadm adjust` to
do a disconnect/connect cycle. This affected `lb-tcp` and `rdma`. As
`tcp` is the default, it was not affected by this.
---
drbd/drbd_nl.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/drbd/drbd_nl.c b/drbd/drbd_nl.c
index 8c968cf252ca..b0a1e6fa46f1 100644
--- a/drbd/drbd_nl.c
+++ b/drbd/drbd_nl.c
@@ -3729,6 +3729,11 @@ int drbd_adm_net_opts(struct sk_buff *skb, struct genl_info *info)
if (should_set_defaults(info))
set_net_conf_defaults(new_net_conf);
+ /* The transport_name is immutable taking precedence over set_net_conf_defaults() */
+ memcpy(new_net_conf->transport_name, old_net_conf->transport_name,
+ old_net_conf->transport_name_len);
+ new_net_conf->transport_name_len = old_net_conf->transport_name_len;
+
err = net_conf_from_attrs_for_change(new_net_conf, info);
if (err && err != -ENOMSG) {
retcode = ERR_MANDATORY_TAG;
--
2.35.3

35
0017-drbd-propagate-exposed-UUIDs-only-into-established-c.patch Normal file
View File

@ -0,0 +1,35 @@
From 0666bc5b014049b6aa184b5a8fdb60481b9c8717 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Fri, 22 Sep 2023 15:09:07 +0200
Subject: [PATCH 17/20] drbd: propagate exposed UUIDs only into established
connections
Sending into connections in C_CONNECTING state while we exchange the
feature and auth packets might interfere with exchanging those. I also
witnessed that it might block unexpectedly long (2 - 6 seconds),
holding the `connection->mutex[DATA_STREAM]` mutex and again causing
troubles with establishing connections.
This is a fix for commit 772e5b21d from April 2023
"drbd: Consider outdating a disk when more recent data is behind a diskless"
Released with of 9.1.15 and 9.2.4.
---
drbd/drbd_receiver.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drbd/drbd_receiver.c b/drbd/drbd_receiver.c
index 0c3ab0fd486c..bfd265b86b43 100644
--- a/drbd/drbd_receiver.c
+++ b/drbd/drbd_receiver.c
@@ -7206,7 +7206,7 @@ static void propagate_exposed_uuid(struct drbd_device *device)
u64 im;
for_each_peer_device_ref(peer_device, im, device) {
- if (peer_device->connection->cstate[NOW] < C_CONNECTING)
+ if (!test_bit(INITIAL_STATE_SENT, &peer_device->flags))
continue;
drbd_send_current_uuid(peer_device, device->exposed_data_uuid, 0);
}
--
2.35.3

185
0018-drbd-rework-autopromote.patch Normal file
View File

@ -0,0 +1,185 @@
From e3ef0e229a6ae88346164d1507697ae1b397cdf9 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Moritz=20=22WanzenBug=22=20Wanzenb=C3=B6ck?=
<moritz.wanzenboeck@linbit.com>
Date: Tue, 12 Sep 2023 10:17:08 +0200
Subject: [PATCH 18/20] drbd: rework autopromote
With upstream commit ae220766d87c we lost the ability to keep separate
counts for RW and RO openers. Instead, we keep track of openers using a
single count, and a flag indicating if the device was opened RW once.
Once a device was opened RW, it will stay "writable" for DRBD, until all
openers are gone. This should offer a good compromise between keeping
the old auto-promote behaviour, and the changed device interface.
---
drbd/drbd_int.h | 3 ++-
drbd/drbd_main.c | 38 +++++++++++++++++++++-----------------
drbd/drbd_nl.c | 9 +++++----
drbd/drbd_state.c | 7 ++++---
4 files changed, 32 insertions(+), 25 deletions(-)
diff --git a/drbd/drbd_int.h b/drbd/drbd_int.h
index b705f26e71a4..cf593c09cda1 100644
--- a/drbd/drbd_int.h
+++ b/drbd/drbd_int.h
@@ -1399,7 +1399,8 @@ struct drbd_device {
struct drbd_bitmap *bitmap;
- int open_rw_cnt, open_ro_cnt;
+ int open_cnt;
+ bool writable;
/* FIXME clean comments, restructure so it is more obvious which
* members are protected by what */
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index 4b0b967c2c97..bb05b2215dfb 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -2579,10 +2579,9 @@ static enum ioc_rv inc_open_count(struct drbd_device *device, fmode_t mode)
r = IOC_ABORT;
else if (!resource->remote_state_change) {
r = IOC_OK;
+ device->open_cnt++;
if (mode & FMODE_WRITE)
- device->open_rw_cnt++;
- else
- device->open_ro_cnt++;
+ device->writable = true;
}
read_unlock_irq(&resource->state_rwlock);
@@ -2756,8 +2755,10 @@ void drbd_open_counts(struct drbd_resource *resource, int *rw_count_ptr, int *ro
rcu_read_lock();
idr_for_each_entry(&resource->devices, device, vnr) {
- rw_count += device->open_rw_cnt;
- ro_count += device->open_ro_cnt;
+ if (device->writable)
+ rw_count += device->open_cnt;
+ else
+ ro_count += device->open_cnt;
}
rcu_read_unlock();
*rw_count_ptr = rw_count;
@@ -2825,33 +2826,35 @@ static void drbd_release(struct gendisk *gd, fmode_t mode)
{
struct drbd_device *device = gd->private_data;
struct drbd_resource *resource = device->resource;
+ bool was_writable = device->writable;
int open_rw_cnt, open_ro_cnt;
mutex_lock(&resource->open_release);
- if (mode & FMODE_WRITE)
- device->open_rw_cnt--;
- else
- device->open_ro_cnt--;
+ device->open_cnt--;
drbd_open_counts(resource, &open_rw_cnt, &open_ro_cnt);
- /* last one to close will be responsible for write-out of all dirty pages */
- if (mode & FMODE_WRITE && device->open_rw_cnt == 0)
+ /* last one to close will be responsible for write-out of all dirty pages.
+ * We also reset the writable flag for this device here: later code may
+ * check if the device is still opened for writes to determine things
+ * like auto-demote.
+ */
+ if (was_writable && device->open_cnt == 0) {
drbd_fsync_device(device);
+ device->writable = false;
+ }
if (open_ro_cnt == 0)
wake_up_all(&resource->state_wait);
- if (test_bit(UNREGISTERED, &device->flags) &&
- device->open_rw_cnt == 0 && device->open_ro_cnt == 0 &&
+ if (test_bit(UNREGISTERED, &device->flags) && device->open_cnt == 0 &&
!test_and_set_bit(DESTROYING_DEV, &device->flags))
call_rcu(&device->rcu, drbd_reclaim_device);
if (resource->res_opts.auto_promote) {
enum drbd_state_rv rv;
- if (mode & FMODE_WRITE &&
- open_rw_cnt == 0 &&
+ if (was_writable && open_rw_cnt == 0 &&
resource->role[NOW] == R_PRIMARY &&
!test_bit(EXPLICIT_PRIMARY, &resource->flags)) {
rv = drbd_set_role(resource, R_SECONDARY, false, "auto-demote", NULL);
@@ -2869,9 +2872,10 @@ static void drbd_release(struct gendisk *gd, fmode_t mode)
end_state_change(resource, &irq_flags, "release");
}
- /* if the open counts are 0, we free the whole list, otherwise we remove the specific pid */
+ /* if the open count is 0, we free the whole list, otherwise we remove the specific pid */
prune_or_free_openers(device,
- (open_ro_cnt == 0 && open_rw_cnt == 0) ? 0 : task_pid_nr(current));
+ (open_ro_cnt == 0 && open_rw_cnt == 0) ?
+ 0 : task_pid_nr(current));
mutex_unlock(&resource->open_release);
diff --git a/drbd/drbd_nl.c b/drbd/drbd_nl.c
index b0a1e6fa46f1..71ed4272614e 100644
--- a/drbd/drbd_nl.c
+++ b/drbd/drbd_nl.c
@@ -4541,8 +4541,10 @@ int drbd_open_ro_count(struct drbd_resource *resource)
int vnr, open_ro_cnt = 0;
read_lock_irq(&resource->state_rwlock);
- idr_for_each_entry(&resource->devices, device, vnr)
- open_ro_cnt += device->open_ro_cnt;
+ idr_for_each_entry(&resource->devices, device, vnr) {
+ if (!device->writable)
+ open_ro_cnt += device->open_cnt;
+ }
read_unlock_irq(&resource->state_rwlock);
return open_ro_cnt;
@@ -6394,8 +6396,7 @@ static enum drbd_ret_code adm_del_minor(struct drbd_device *device)
notify_device_state(NULL, 0, device, NULL, NOTIFY_DESTROY);
mutex_unlock(&notification_mutex);
- if (device->open_ro_cnt == 0 && device->open_rw_cnt == 0 &&
- !test_and_set_bit(DESTROYING_DEV, &device->flags))
+ if (device->open_cnt == 0 && !test_and_set_bit(DESTROYING_DEV, &device->flags))
call_rcu(&device->rcu, drbd_reclaim_device);
return ret;
diff --git a/drbd/drbd_state.c b/drbd/drbd_state.c
index 22cd134be962..fa70507df425 100644
--- a/drbd/drbd_state.c
+++ b/drbd/drbd_state.c
@@ -1634,7 +1634,7 @@ handshake_found:
return SS_TWO_PRIMARIES;
if (!fail_io[NEW]) {
idr_for_each_entry(&resource->devices, device, vnr) {
- if (device->open_ro_cnt)
+ if (!device->writable && device->open_cnt)
return SS_PRIMARY_READER;
/*
* One might be tempted to add "|| open_rw_cont" here.
@@ -1661,7 +1661,7 @@ handshake_found:
(disk_state[OLD] > D_DETACHING && disk_state[NEW] == D_DETACHING)))
return SS_IN_TRANSIENT_STATE;
- if (role[OLD] == R_PRIMARY && role[NEW] == R_SECONDARY && device->open_rw_cnt &&
+ if (role[OLD] == R_PRIMARY && role[NEW] == R_SECONDARY && device->writable &&
!(resource->state_change_flags & CS_FS_IGN_OPENERS))
return SS_DEVICE_IN_USE;
@@ -1693,7 +1693,8 @@ handshake_found:
return SS_NO_UP_TO_DATE_DISK;
/* Prevent detach or disconnect while held open read only */
- if (device->open_ro_cnt && any_disk_up_to_date[OLD] && !any_disk_up_to_date[NEW])
+ if (!device->writable && device->open_cnt &&
+ any_disk_up_to_date[OLD] && !any_disk_up_to_date[NEW])
return SS_NO_UP_TO_DATE_DISK;
if (disk_state[NEW] == D_NEGOTIATING)
--
2.35.3

107
0019-compat-block-remove-the-unused-mode-argument-to-rele.patch Normal file
View File

@ -0,0 +1,107 @@
From bf287c15359c4495a706ca270e7dcab4c2d6ebcc Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Moritz=20=22WanzenBug=22=20Wanzenb=C3=B6ck?=
<moritz.wanzenboeck@linbit.com>
Date: Tue, 12 Sep 2023 15:27:01 +0200
Subject: [PATCH 19/20] compat: block: remove the unused mode argument to
->release
See also upstream Linux kernel commit
ae220766d87c ("block: remove the unused mode argument to ->release")
Original message:
The mode argument to the ->release block_device_operation is never used,
so remove it.
---
...ions_release__no_takes_single_argument.cocci | 17 +++++++++++++++++
drbd/drbd-kernel-compat/gen_patch_names.c | 3 +++
...e_operations_release_takes_single_argument.c | 7 +++++++
drbd/drbd_main.c | 6 +++---
4 files changed, 30 insertions(+), 3 deletions(-)
create mode 100644 drbd/drbd-kernel-compat/cocci/block_device_operations_release__no_takes_single_argument.cocci
create mode 100644 drbd/drbd-kernel-compat/tests/block_device_operations_release_takes_single_argument.c
diff --git a/drbd/drbd-kernel-compat/cocci/block_device_operations_release__no_takes_single_argument.cocci b/drbd/drbd-kernel-compat/cocci/block_device_operations_release__no_takes_single_argument.cocci
new file mode 100644
index 000000000000..b0a0ee9eef39
--- /dev/null
+++ b/drbd/drbd-kernel-compat/cocci/block_device_operations_release__no_takes_single_argument.cocci
@@ -0,0 +1,17 @@
+@@
+identifier gd;
+fresh identifier mode = "" ## "mode";
+@@
+ drbd_release(
+ struct gendisk *gd
++ , fmode_t mode
+ ) { ... }
+
+@@
+symbol drbd_release;
+expression gd;
+@@
+ drbd_release(
+ gd
++ , 0
+ )
diff --git a/drbd/drbd-kernel-compat/gen_patch_names.c b/drbd/drbd-kernel-compat/gen_patch_names.c
index 4761c1ef7d0c..608be8b4f099 100644
--- a/drbd/drbd-kernel-compat/gen_patch_names.c
+++ b/drbd/drbd-kernel-compat/gen_patch_names.c
@@ -565,6 +565,9 @@ int main(int argc, char **argv)
patch(1, "blkdev_put", true, false,
COMPAT_BLKDEV_PUT_HAS_HOLDER, "has_holder");
+ patch(1, "block_device_operations_release", true, false,
+ COMPAT_BLOCK_DEVICE_OPERATIONS_RELEASE_TAKES_SINGLE_ARGUMENT, "takes_single_argument");
+
/* #define BLKDEV_ISSUE_ZEROOUT_EXPORTED */
/* #define BLKDEV_ZERO_NOUNMAP */
diff --git a/drbd/drbd-kernel-compat/tests/block_device_operations_release_takes_single_argument.c b/drbd/drbd-kernel-compat/tests/block_device_operations_release_takes_single_argument.c
new file mode 100644
index 000000000000..d51c863e359a
--- /dev/null
+++ b/drbd/drbd-kernel-compat/tests/block_device_operations_release_takes_single_argument.c
@@ -0,0 +1,7 @@
+/* { "version": "v6.5-rc1", "commit": "ae220766d87cd6799dbf918fea10613ae14c0654", "comment": "block: remove the unused mode argument to ->release", "author": "Christoph Hellwig <hch@lst.de>", "date": "Thu Jun 8 13:02:37 2023 +0200" } */
+#include <linux/blkdev.h>
+
+void foo(struct block_device_operations *ops, struct gendisk *gd)
+{
+ ops->release(gd);
+}
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index bb05b2215dfb..1864861db21d 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -53,7 +53,7 @@
#include "drbd_dax_pmem.h"
static int drbd_open(struct gendisk *gd, fmode_t mode);
-static void drbd_release(struct gendisk *gd, fmode_t mode);
+static void drbd_release(struct gendisk *gd);
static void md_sync_timer_fn(struct timer_list *t);
static int w_bitmap_io(struct drbd_work *w, int unused);
static int flush_send_buffer(struct drbd_connection *connection, enum drbd_stream drbd_stream);
@@ -2740,7 +2740,7 @@ out:
mutex_unlock(&resource->open_release);
if (err) {
- drbd_release(gd, mode);
+ drbd_release(gd);
if (err == -EAGAIN && !(mode & FMODE_NDELAY))
err = -EMEDIUMTYPE;
}
@@ -2822,7 +2822,7 @@ void drbd_fsync_device(struct drbd_device *device)
drbd_flush_peer_acks(resource);
}
-static void drbd_release(struct gendisk *gd, fmode_t mode)
+static void drbd_release(struct gendisk *gd)
{
struct drbd_device *device = gd->private_data;
struct drbd_resource *resource = device->resource;
--
2.35.3

66
0020-drbd-do-not-allow-auto-demote-to-be-interrupted-by-s.patch Normal file
View File

@ -0,0 +1,66 @@
From c8e2a3c4355b4794267cd6e58a074802b4607cb9 Mon Sep 17 00:00:00 2001
From: Joel Colledge <joel.colledge@linbit.com>
Date: Fri, 22 Sep 2023 16:57:24 +0200
Subject: [PATCH 20/20] drbd: do not allow auto-demote to be interrupted by
signal
Pending signals can mess up auto-demote:
drbd res: Preparing cluster-wide state change 671410162 (0->-1 3/2)
drbd res: Aborting cluster-wide state change 671410162 (6ms) rv = -21
drbd res: Auto-demote failed: Interrupted state change
After this state change failure no process has the DRBD device open, but
DRBD remains Primary.
---
drbd/drbd_main.c | 33 +++++++++++++++++++++++----------
1 file changed, 23 insertions(+), 10 deletions(-)
diff --git a/drbd/drbd_main.c b/drbd/drbd_main.c
index 1864861db21d..0719229f210e 100644
--- a/drbd/drbd_main.c
+++ b/drbd/drbd_main.c
@@ -2851,17 +2851,30 @@ static void drbd_release(struct gendisk *gd)
!test_and_set_bit(DESTROYING_DEV, &device->flags))
call_rcu(&device->rcu, drbd_reclaim_device);
- if (resource->res_opts.auto_promote) {
- enum drbd_state_rv rv;
+ if (resource->res_opts.auto_promote &&
+ open_rw_cnt == 0 &&
+ resource->role[NOW] == R_PRIMARY &&
+ !test_bit(EXPLICIT_PRIMARY, &resource->flags)) {
+ sigset_t mask, oldmask;
+ int rv;
+
+ /*
+ * Auto-demote is triggered by the last opener releasing the
+ * DRBD device. However, it is an implicit action, so it should
+ * not be affected by the state of the process. In particular,
+ * it should ignore any pending signals. It may be the case
+ * that the process is releasing DRBD because it is being
+ * terminated using a signal.
+ */
+ sigfillset(&mask);
+ sigprocmask(SIG_BLOCK, &mask, &oldmask);
- if (was_writable && open_rw_cnt == 0 &&
- resource->role[NOW] == R_PRIMARY &&
- !test_bit(EXPLICIT_PRIMARY, &resource->flags)) {
- rv = drbd_set_role(resource, R_SECONDARY, false, "auto-demote", NULL);
- if (rv < SS_SUCCESS)
- drbd_warn(resource, "Auto-demote failed: %s (%d)\n",
- drbd_set_st_err_str(rv), rv);
- }
+ rv = drbd_set_role(resource, R_SECONDARY, false, "auto-demote", NULL);
+ if (rv < SS_SUCCESS)
+ drbd_warn(resource, "Auto-demote failed: %s (%d)\n",
+ drbd_set_st_err_str(rv), rv);
+
+ sigprocmask(SIG_SETMASK, &oldmask, NULL);
}
if (open_ro_cnt == 0 && open_rw_cnt == 0 && resource->fail_io[NOW]) {
--
2.35.3

86
bsc-1215699_fix-build-error-against-kernel-v6.5.4.patch Normal file
View File

@ -0,0 +1,86 @@
Author: Heming Zhao <heming.zhao@suse.com>
Date: Thu Sep 28 11:12:00 2023 +0800
Subject: drbd: fix build error against kernel v6.5.4
Below two upstream commits broken drbd v9.1 building
(belonging to v6.5-rc1)
```
commit dc97391e661009eab46783030d2404c9b6e6f2e7
Author: David Howells <dhowells@redhat.com>
Date: Fri Jun 23 23:55:12 2023 +0100
sock: Remove ->sendpage*() in favour of sendmsg(MSG_SPLICE_PAGES)
commit 0733ad8002916b9dbbbcfe6e92ad44d2657de1c1
Author: Christoph Hellwig <hch@lst.de>
Date: Thu Jun 8 13:02:58 2023 +0200
fs: remove the now unused FMODE_* flags
```
This patch fix the build error.
---
diff -Nupr a/drbd/drbd_main.c b/drbd/drbd_main.c
--- a/drbd/drbd_main.c 2023-09-26 11:08:57.142974614 +0800
+++ b/drbd/drbd_main.c 2023-09-26 11:07:53.646071680 +0800
@@ -2569,7 +2569,7 @@ enum ioc_rv {
static enum ioc_rv inc_open_count(struct drbd_device *device, fmode_t mode)
{
struct drbd_resource *resource = device->resource;
- enum ioc_rv r = mode & FMODE_NDELAY ? IOC_ABORT : IOC_SLEEP;
+ enum ioc_rv r = mode & BLK_OPEN_NDELAY ? IOC_ABORT : IOC_SLEEP;
if (test_bit(DOWN_IN_PROGRESS, &resource->flags))
return IOC_ABORT;
@@ -2695,12 +2695,12 @@ static int drbd_open(struct gendisk *gd,
if (mode & FMODE_WRITE) {
if (resource->role[NOW] == R_SECONDARY) {
- rv = try_to_promote(device, timeout, (mode & FMODE_NDELAY));
+ rv = try_to_promote(device, timeout, (mode & BLK_OPEN_NDELAY));
if (rv < SS_SUCCESS)
drbd_info(resource, "Auto-promote failed: %s (%d)\n",
drbd_set_st_err_str(rv), rv);
}
- } else if ((mode & FMODE_NDELAY) == 0) {
+ } else if ((mode & BLK_OPEN_NDELAY) == 0) {
/* Double check peers
*
* Some services may try to first open ro, and only if that
@@ -2741,7 +2741,7 @@ out:
mutex_unlock(&resource->open_release);
if (err) {
drbd_release(gd);
- if (err == -EAGAIN && !(mode & FMODE_NDELAY))
+ if (err == -EAGAIN && !(mode & BLK_OPEN_NDELAY))
err = -EMEDIUMTYPE;
}
diff -Nupr a/drbd/drbd_transport_tcp.c b/drbd/drbd_transport_tcp.c
--- a/drbd/drbd_transport_tcp.c 2023-09-26 11:09:04.016333505 +0800
+++ b/drbd/drbd_transport_tcp.c 2023-09-26 11:07:59.132758771 +0800
@@ -1193,6 +1193,8 @@ static int dtt_send_page(struct drbd_tra
struct drbd_tcp_transport *tcp_transport =
container_of(transport, struct drbd_tcp_transport, transport);
struct socket *socket = tcp_transport->stream[stream];
+ struct msghdr msghdr = { .msg_flags = MSG_SPLICE_PAGES|msg_flags, };
+ struct bio_vec bvec;
int len = size;
int err = -EIO;
@@ -1204,7 +1206,10 @@ static int dtt_send_page(struct drbd_tra
do {
int sent;
- sent = socket->ops->sendpage(socket, page, offset, len, msg_flags);
+ //sent = socket->ops->sendpage(socket, page, offset, len, msg_flags);
+ bvec_set_page(&bvec, page, offset, len);
+ iov_iter_bvec(&msghdr.msg_iter, ITER_SOURCE, &bvec, 1, len);
+ sent = sock_sendmsg(socket, &msghdr);
if (sent <= 0) {
if (sent == -EAGAIN) {
if (drbd_stream_send_timed_out(transport, stream))

28
drbd.changes
View File

@ -1,3 +1,31 @@
------------------------------------------------------------------
Tue Sep 26 03:35:00 UTC 2023 - Heming Zhao <heming.zhao@suse.com>
- drbd: fix build error against kernel v6.5.4 (boo#1215699)
* add upstream patch
+ 0001-drbd-allow-transports-to-take-additional-krefs-on-a-.patch
+ 0002-drbd-improve-decision-about-marking-a-failed-disk-Ou.patch
+ 0003-drbd-fix-error-path-in-drbd_get_listener.patch
+ 0004-drbd-build-fix-spurious-re-build-attempt-of-compat.p.patch
+ 0005-drbd-log-error-code-when-thread-fails-to-start.patch
+ 0006-drbd-log-numeric-value-of-drbd_state_rv-as-well-as-s.patch
+ 0007-drbd-stop-defining-__KERNEL_SYSCALLS__.patch
+ 0008-compat-block-introduce-holder-ops.patch
+ 0009-drbd-reduce-net_ee-not-empty-info-to-a-dynamic-debug.patch
+ 0010-drbd-do-not-send-P_CURRENT_UUID-to-DRBD-8-peer-when-.patch
+ 0011-compat-block-pass-a-gendisk-to-open.patch
+ 0012-drbd-Restore-DATA_CORKED-and-CONTROL_CORKED-bits.patch
+ 0013-drbd-remove-unused-extern-for-conn_try_outdate_peer.patch
+ 0014-drbd-include-source-of-state-change-in-log.patch
+ 0015-compat-block-use-the-holder-as-indication-for-exclus.patch
+ 0016-drbd-Fix-net-options-set-defaults-to-not-clear-the-t.patch
+ 0017-drbd-propagate-exposed-UUIDs-only-into-established-c.patch
+ 0018-drbd-rework-autopromote.patch
+ 0019-compat-block-remove-the-unused-mode-argument-to-rele.patch
+ 0020-drbd-do-not-allow-auto-demote-to-be-interrupted-by-s.patch
* add suse special patch
+ bsc-1215699_fix-build-error-against-kernel-v6.5.4.patch
------------------------------------------------------------------
Mon Sep 04 11:35:00 UTC 2023 - Heming Zhao <heming.zhao@suse.com>

54
drbd.spec
View File

@ -34,8 +34,33 @@ Source1: preamble
Source2: Module.supported
Source3: drbd_git_revision
Patch1: bsc-1025089_fix-resync-finished-with-syncs-have-bits-set.patch
Patch99: suse-coccinelle.patch
########################
# upstream patch
Patch0001: 0001-drbd-allow-transports-to-take-additional-krefs-on-a-.patch
Patch0002: 0002-drbd-improve-decision-about-marking-a-failed-disk-Ou.patch
Patch0003: 0003-drbd-fix-error-path-in-drbd_get_listener.patch
Patch0004: 0004-drbd-build-fix-spurious-re-build-attempt-of-compat.p.patch
Patch0005: 0005-drbd-log-error-code-when-thread-fails-to-start.patch
Patch0006: 0006-drbd-log-numeric-value-of-drbd_state_rv-as-well-as-s.patch
Patch0007: 0007-drbd-stop-defining-__KERNEL_SYSCALLS__.patch
Patch0008: 0008-compat-block-introduce-holder-ops.patch
Patch0009: 0009-drbd-reduce-net_ee-not-empty-info-to-a-dynamic-debug.patch
Patch0010: 0010-drbd-do-not-send-P_CURRENT_UUID-to-DRBD-8-peer-when-.patch
Patch0011: 0011-compat-block-pass-a-gendisk-to-open.patch
Patch0012: 0012-drbd-Restore-DATA_CORKED-and-CONTROL_CORKED-bits.patch
Patch0013: 0013-drbd-remove-unused-extern-for-conn_try_outdate_peer.patch
Patch0014: 0014-drbd-include-source-of-state-change-in-log.patch
Patch0015: 0015-compat-block-use-the-holder-as-indication-for-exclus.patch
Patch0016: 0016-drbd-Fix-net-options-set-defaults-to-not-clear-the-t.patch
Patch0017: 0017-drbd-propagate-exposed-UUIDs-only-into-established-c.patch
Patch0018: 0018-drbd-rework-autopromote.patch
Patch0019: 0019-compat-block-remove-the-unused-mode-argument-to-rele.patch
Patch0020: 0020-drbd-do-not-allow-auto-demote-to-be-interrupted-by-s.patch
# suse special patch
Patch1001: bsc-1025089_fix-resync-finished-with-syncs-have-bits-set.patch
Patch1002: suse-coccinelle.patch
Patch1003: bsc-1215699_fix-build-error-against-kernel-v6.5.4.patch
########################
#https://github.com/openSUSE/rpmlint-checks/blob/master/KMPPolicyCheck.py
BuildRequires: coccinelle >= 1.0.8
@ -70,8 +95,29 @@ installed kernel.
%prep
%setup -q -n drbd-%{version}
%patch1 -p1
%patch99 -p1
%patch0001 -p1
%patch0002 -p1
%patch0003 -p1
%patch0004 -p1
%patch0005 -p1
%patch0006 -p1
%patch0007 -p1
%patch0008 -p1
%patch0009 -p1
%patch0010 -p1
%patch0011 -p1
%patch0012 -p1
%patch0013 -p1
%patch0014 -p1
%patch0015 -p1
%patch0016 -p1
%patch0017 -p1
%patch0018 -p1
%patch0019 -p1
%patch0020 -p1
%patch1001 -p1
%patch1002 -p1
%patch1003 -p1
mkdir source
cp -a drbd/. source/. || :