[Cluster-devel] [PATCHv2 RESEND v5.10-rc2 02/13] fs: dlm: define max send buffer

[Alexander Aring <aahringo@redhat.com>]

This patch will set the maximum transmit buffer size for rcom messages
with "names" to 4096 bytes. It's a leftover change of
commit 4798cbbfbd00 ("fs: dlm: rework receive handling"). Fact is that we
cannot allocate a contiguous transmit buffer length above of 4096 bytes.
It seems at some places the upper layer protocol will calculate according
to dlm_config.ci_buffer_size the possible payload of a dlm recovery
message. As compiler setting we will use now the maximum possible
message which dlm can send out. Commit 4e192ee68e5af ("fs: dlm: disallow
buffer size below default") disallow a buffer setting smaller than the
4096 bytes and above 4096 bytes is definitely wrong because we will then
write out of buffer space as we cannot allocate a contiguous buffer above
4096 bytes. The ci_buffer_size is still there to define the possible
maximum receive buffer size of a recvmsg() which should be at least the
maximum possible dlm message size.

Signed-off-by: Alexander Aring <aahringo@redhat.com>
---
 fs/dlm/lockspace.c | 2 +-
 fs/dlm/lowcomms.h  | 2 ++
 fs/dlm/member.c    | 2 +-
 fs/dlm/rcom.c      | 6 +++---
 4 files changed, 7 insertions(+), 5 deletions(-)

diff --git a/fs/dlm/lockspace.c b/fs/dlm/lockspace.c
index 624617c12250..561dcad08ad6 100644
--- a/fs/dlm/lockspace.c
+++ b/fs/dlm/lockspace.c
@@ -572,7 +572,7 @@ static int new_lockspace(const char *name, const char *cluster,
 	mutex_init(&ls->ls_requestqueue_mutex);
 	mutex_init(&ls->ls_clear_proc_locks);
 
-	ls->ls_recover_buf = kmalloc(dlm_config.ci_buffer_size, GFP_NOFS);
+	ls->ls_recover_buf = kmalloc(LOWCOMMS_MAX_TX_BUFFER_LEN, GFP_NOFS);
 	if (!ls->ls_recover_buf)
 		goto out_lkbidr;
 
diff --git a/fs/dlm/lowcomms.h b/fs/dlm/lowcomms.h
index 687b2894e469..0918f9376489 100644
--- a/fs/dlm/lowcomms.h
+++ b/fs/dlm/lowcomms.h
@@ -12,6 +12,8 @@
 #ifndef __LOWCOMMS_DOT_H__
 #define __LOWCOMMS_DOT_H__
 
+#define LOWCOMMS_MAX_TX_BUFFER_LEN	4096
+
 int dlm_lowcomms_start(void);
 void dlm_lowcomms_stop(void);
 void dlm_lowcomms_exit(void);
diff --git a/fs/dlm/member.c b/fs/dlm/member.c
index 7ad83deb4505..ceef3f2074ff 100644
--- a/fs/dlm/member.c
+++ b/fs/dlm/member.c
@@ -270,7 +270,7 @@ int dlm_slots_assign(struct dlm_ls *ls, int *num_slots, int *slots_size,
 
 	log_slots(ls, gen, num, NULL, array, array_size);
 
-	max_slots = (dlm_config.ci_buffer_size - sizeof(struct dlm_rcom) -
+	max_slots = (LOWCOMMS_MAX_TX_BUFFER_LEN - sizeof(struct dlm_rcom) -
 		     sizeof(struct rcom_config)) / sizeof(struct rcom_slot);
 
 	if (num > max_slots) {
diff --git a/fs/dlm/rcom.c b/fs/dlm/rcom.c
index 4daf5dc2b51c..73ddee5159d7 100644
--- a/fs/dlm/rcom.c
+++ b/fs/dlm/rcom.c
@@ -162,7 +162,7 @@ int dlm_rcom_status(struct dlm_ls *ls, int nodeid, uint32_t status_flags)
 	set_rcom_status(ls, (struct rcom_status *)rc->rc_buf, status_flags);
 
 	allow_sync_reply(ls, &rc->rc_id);
-	memset(ls->ls_recover_buf, 0, dlm_config.ci_buffer_size);
+	memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
 
 	send_rcom(ls, mh, rc);
 
@@ -284,7 +284,7 @@ int dlm_rcom_names(struct dlm_ls *ls, int nodeid, char *last_name, int last_len)
 	memcpy(rc->rc_buf, last_name, last_len);
 
 	allow_sync_reply(ls, &rc->rc_id);
-	memset(ls->ls_recover_buf, 0, dlm_config.ci_buffer_size);
+	memset(ls->ls_recover_buf, 0, LOWCOMMS_MAX_TX_BUFFER_LEN);
 
 	send_rcom(ls, mh, rc);
 
@@ -304,7 +304,7 @@ static void receive_rcom_names(struct dlm_ls *ls, struct dlm_rcom *rc_in)
 
 	nodeid = rc_in->rc_header.h_nodeid;
 	inlen = rc_in->rc_header.h_length - sizeof(struct dlm_rcom);
-	outlen = dlm_config.ci_buffer_size - sizeof(struct dlm_rcom);
+	outlen = LOWCOMMS_MAX_TX_BUFFER_LEN - sizeof(struct dlm_rcom);
 
 	error = create_rcom(ls, nodeid, DLM_RCOM_NAMES_REPLY, outlen, &rc, &mh);
 	if (error)
-- 
2.26.2