fs: change ast and bast trace order
Subsystems (2)
Alexander Aring
maintainerPATCH 1/3 fs: change ast and bast trace order
This patch changes the order to call trace functionality before calling the traced callback. The intention is always to see at first that a dlm callback occurred and then optionally see dlm user traces in the ast or bast callback. Currently the behaviour is vice versa, the user sees that dlm ast or bast callback occurred after the dlm user callback for ast or bast was called.
--- fs/dlm/ast.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
fs/dlm/ast.c +2 -2
@@ -255,13 +255,13 @@ void dlm_callback_work(struct work_struct *work)if (callbacks[i].flags &DLM_CB_SKIP ) {continue ; }else if (callbacks[i].flags &DLM_CB_BAST ) { -bastfn (lkb->lkb_astparam, callbacks[i].mode);trace_dlm_bast (ls, lkb, callbacks[i].mode); +bastfn (lkb->lkb_astparam, callbacks[i].mode); }else if (callbacks[i].flags &DLM_CB_CAST ) { lkb->lkb_lksb->sb_status = callbacks[i].sb_status; lkb->lkb_lksb->sb_flags = callbacks[i].sb_flags; -castfn (lkb->lkb_astparam);trace_dlm_ast (ls, lkb, lkb->lkb_lksb); +castfn (lkb->lkb_astparam); } }
Signature · 1 line
2.31.1
Alexander Aring
maintainerPATCH 2/3 fs: remove additional dereference of lkbsb
This patch removes a dereference of lkbsb of lkb when calling ast tracepoint. First it reduces additional overhead, even if traces are not acitivated. Second we can deference it in TP_fast_assign over the already passed lkb parameter.
--- fs/dlm/ast.c | 2 +- include/trace/events/dlm.h | 8 ++++---- 2 files changed, 5 insertions(+), 5 deletions(-)
fs/dlm/ast.c +1 -1
@@ -260,7 +260,7 @@ void dlm_callback_work(struct work_struct *work) }else if (callbacks[i].flags &DLM_CB_CAST ) { lkb->lkb_lksb->sb_status = callbacks[i].sb_status; lkb->lkb_lksb->sb_flags = callbacks[i].sb_flags; -trace_dlm_ast (ls, lkb, lkb->lkb_lksb); +trace_dlm_ast (ls, lkb);castfn (lkb->lkb_astparam); } }
include/trace/events/dlm.h +4 -4
@@ -138,9 +138,9 @@ TRACE_EVENT(dlm_bast, TRACE_EVENT(dlm_ast, - TP_PROTO(struct dlm_ls *ls, struct dlm_lkb *lkb, struct dlm_lksb *lksb), + TP_PROTO(struct dlm_ls *ls, struct dlm_lkb *lkb), - TP_ARGS(ls, lkb, lksb), + TP_ARGS(ls, lkb), TP_STRUCT__entry( __field(__u32, ls_id)
@@ -152,8 +152,8 @@ TRACE_EVENT(dlm_ast,TP_fast_assign ( __entry->ls_id = ls->ls_global_id; __entry->lkb_id = lkb->lkb_id; - __entry->sb_flags = lksb->sb_flags; - __entry->sb_status = lksb->sb_status; + __entry->sb_flags = lkb->lkb_lksb->sb_flags; + __entry->sb_status = lkb->lkb_lksb->sb_status; ), TP_printk("ls_id=%u lkb_id=%x sb_flags=%s sb_status=%d",
Signature · 1 line
2.31.1
Alexander Aring
maintainerPATCH 3/3 fs: dlm: add resource name to tracepoints
In reply to Alexander AringThis patch adds the resource name to dlm tracepoints. In case of dlm_lock() we might end in a short time situation (if flags is not convert) that a lkb is not associated with a resource at the time when the tracepoint is called. Therefore we have a a prioritzation of getting the resource name. If the resource in a lkb isn't set we use the name and namelen passed as parameter as fallback.
The dlm resource name can be decoded as a string representated ascii codec. DLM itself stores the name with as a null terminated array but the user does not required to pass a null terminated resource name. To have somehow a similar behaviour and the user knows that the resource name of the dlm user is ascii codec we store the resource name array in a null terminated array but pass the array length for the trace user without the null terminated byte. The advantage here is that the user can run directly a printf() on the resource name array with a optional check on isascii().
The TP_printk() call however use always a hexadecimal string representation for the resource name array.
--- fs/dlm/lock.c | 4 +- include/trace/events/dlm.h | 134 ++++++++++++++++++++++++++++++++----- 2 files changed, 118 insertions(+), 20 deletions(-)
fs/dlm/lock.c +2 -2
@@ -3472,7 +3472,7 @@ int dlm_lock(dlm_lockspace_t *lockspace,if (error) goto out; -trace_dlm_lock_start (ls, lkb, mode, flags); +trace_dlm_lock_start (ls, lkb, name, namelen, mode, flags); error =set_lock_args (mode, lksb, flags, namelen,0 , ast, astarg, bast, &args);
@@ -3487,7 +3487,7 @@ int dlm_lock(dlm_lockspace_t *lockspace,if (error == -EINPROGRESS ) error =0 ; out_put: -trace_dlm_lock_end (ls, lkb, mode, flags, error); +trace_dlm_lock_end (ls, lkb, name, namelen, mode, flags, error);if (convert || error)__put_lkb (ls, lkb);
include/trace/events/dlm.h +116 -18
249 lines not on this page · raw message
Signature · 1 line
2.31.1
Alexander Aring
maintainerPATCH 3/3 fs: dlm: add resource name to tracepoints
Hi,
Quoted 18 lines · Alexander Aring
This patch adds the resource name to dlm tracepoints. In case of
dlm_lock()we might end in a short time situation (if flags is not convert) that a lkb is not associated with a resource at the time when the tracepoint is called. Therefore we have a a prioritzation of getting the resource name. If the resource in a lkb isn't set we use the name and namelen passed as parameter as fallback.The dlm resource name can be decoded as a string representated ascii codec. DLM itself stores the name with as a null terminated array but the user does not required to pass a null terminated resource name. To have somehow a similar behaviour and the user knows that the resource name of the dlm user is ascii codec we store the resource name array in a null terminated array but pass the array length for the trace user without the null terminated byte. The advantage here is that the user can run directly a
printf()on the resource name array with a optional check onisascii().
Quoted 3 lines
The
TP_printk()call however use always a hexadecimal string representation for the resource name array.
I forgot to mention here to say something about the rsb lock and refcount. Regarding the refcount it's stimple, if it's associated with a lkb, it can't be freed at this time... and the rsb lock, we access the name here only and I think this should never be changed.
- Alex
Alexander Aring
maintainerPATCH 3/3 fs: dlm: add resource name to tracepoints
In reply to Alexander AringHi,
Quoted 62 lines · Alexander Aring
This patch adds the resource name to dlm tracepoints. In case of
dlm_lock()we might end in a short time situation (if flags is not convert) that a lkb is not associated with a resource at the time when the tracepoint is called. Therefore we have a a prioritzation of getting the resource name. If the resource in a lkb isn't set we use the name and namelen passed as parameter as fallback.The dlm resource name can be decoded as a string representated ascii codec. DLM itself stores the name with as a null terminated array but the user does not required to pass a null terminated resource name. To have somehow a similar behaviour and the user knows that the resource name of the dlm user is ascii codec we store the resource name array in a null terminated array but pass the array length for the trace user without the null terminated byte. The advantage here is that the user can run directly a
printf()on the resource name array with a optional check onisascii().The
TP_printk()call however use always a hexadecimal string representation for the resource name array.Signed-off-by Alexander Aring <[email protected]>--- fs/dlm/lock.c | 4 +- include/trace/events/dlm.h | 134 ++++++++++++++++++++++++++++++++----- 2 files changed, 118 insertions(+), 20 deletions(-)fs/dlm/lock.c +2
-2@@ -3472,7 +3472,7 @@ int dlm_lock(dlm_lockspace_t *lockspace,if (error) goto out; -trace_dlm_lock_start (ls, lkb, mode, flags); +trace_dlm_lock_start (ls, lkb, name, namelen, mode, flags); error =set_lock_args (mode, lksb, flags, namelen,0 , ast, astarg, bast, &args);@@ -3487,7 +3487,7 @@ int dlm_lock(dlm_lockspace_t *lockspace,if (error == -EINPROGRESS ) error =0 ; out_put: -trace_dlm_lock_end (ls, lkb, mode, flags, error); +trace_dlm_lock_end (ls, lkb, name, namelen, mode, flags, error);if (convert || error)__put_lkb (ls, lkb);include/trace/events/dlm.h +2
-2@@ -49,38 +49,56 @@/* note: we begin tracing dlm_lock_start() only if ls and lkb are found */ TRACE_EVENT(dlm_lock_start, - TP_PROTO(struct dlm_ls *ls, struct dlm_lkb *lkb, int mode, - __u32 flags), + TP_PROTO(struct dlm_ls *ls, struct dlm_lkb *lkb, void *name, + unsigned int namelen, int mode, __u32 flags),
include/trace/events/dlm.h line 63
Quoted 10 lines
include/trace/events/dlm.h +3
-1- TP_ARGS(ls, lkb, mode, flags), + TP_ARGS(ls, lkb, name, namelen, mode, flags),TP_STRUCT__entry (__field (__u32 ,ls_id )__field (__u32, lkb_id)__field (int, mode)__field (__u32, flags) + __dynamic_array(unsigned char, res_name, + lkb->lkb_resource ? lkb->lkb_resource->res_length + 1 : namelen + 1)
I will send a v2 series for this. I think we should remove the additional byte here and let the user handle it to parse it in some way and add a null termination on its own (if necessary). The array length in trace api is encoded in their trace UAPI/data structure, we can't make it ending '\0' and make the array length without the null termination. Or we add another field for the length and have two lengths there, whereas one is -1 of the other... sounds not right.
However..., I will change that again.
The user should not assume to have a '\0' terminated array here, _whereas_ in dlm_rsb kernel the user can assume it... but can't assume it's ascii code.
Quoted 11 lines
include/trace/events/dlm.h +4
-0),TP_fast_assign ( + struct dlm_rsb *r; +char *c; + __entry->ls_id = ls->ls_global_id; __entry->lkb_id = lkb->lkb_id; __entry->mode = mode; __entry->flags = flags; +
include/trace/events/dlm.h line 84
Quoted 10 lines
include/trace/events/dlm.h +10
-0+ r = lkb->lkb_resource; +if (r) +memcpy (__get_dynamic_array (res_name), r->res_name, +__get_dynamic_array_len (res_name)); +else if (name) +memcpy (__get_dynamic_array (res_name), name, +__get_dynamic_array_len (res_name)); + + c =__get_dynamic_array (res_name); + c[__get_dynamic_array_len (res_name) -1 ] ='\0' ;
Hopefully we can remove this then...
include/trace/events/dlm.h line 92
Quoted 10 lines
include/trace/events/dlm.h +4
-2), -TP_printk ("ls_id=%u lkb_id=%x mode=%s flags=%s" , +TP_printk ("ls_id=%u lkb_id=%x mode=%s flags=%s res_name=%s" , __entry->ls_id, __entry->lkb_id,show_lock_mode (__entry->mode), -show_lock_flags (__entry->flags)) +show_lock_flags (__entry->flags), +__print_hex_str (__get_dynamic_array (res_name), +__get_dynamic_array_len (res_name)))
also wrong that it will always print the 00 byte at the end. :-/
- Alex