1PMEXTRACTVALUE(3) Library Functions Manual PMEXTRACTVALUE(3)
2
3
4
6 pmExtractValue - extract a performance metric value from a pmResult
7 structure
8
10 #include <pcp/pmapi.h>
11
12 int pmExtractValue(int valfmt, const pmValue *ival, int itype,
13 pmAtomValue *oval, int otype);
14
15 cc ... -lpcp
16
18 The pmValue structure is embedded within the pmResult structure that is
19 used to return one or more performance metrics; see pmFetch(3).
20
21 All performance metric values may be encoded in a pmAtomValue union,
22 defined as follows;
23
24 typedef union {
25 __int32_t l; /* 32-bit signed */
26 __uint32_t ul; /* 32-bit unsigned */
27 __int64_t ll; /* 64-bit signed */
28 __uint64_t ull; /* 64-bit unsigned */
29 float f; /* 32-bit floating point */
30 double d; /* 64-bit floating point */
31 char *cp; /* char ptr */
32 pmValueBlock *vbp; /* pmValueBlock ptr */
33 } pmAtomValue;
34
35 The routine pmExtractValue provides a convenient mechanism for extract‐
36 ing values from the pmValue part of a pmResult structure, optionally
37 converting the data type, and making the result available to the appli‐
38 cation programmer.
39
40 itype defines the data type of the input value held in ival according
41 to the storage format defined by valfmt (see pmFetch(3)). otype de‐
42 fines the data type of the result to be placed in oval.
43
44 The value for itype is typically extracted from a pmDesc structure,
45 following a call to pmLookupDesc(3) for a particular performance met‐
46 ric.
47
48 The otype value should be one of the defined PM_TYPE_... values, that
49 have a 1:1 correspondence with the fields in the pmAtomValue union.
50
51 Normally the valfmt parameter would be plucked from the same pmResult
52 structure that provides the ival parameter, and if valfmt specifies
53 PM_VAL_INSITU, then the following types are not allowed, as these can‐
54 not be encoded in 32-bits; __int64_t, __uint64_t, double, char * and
55 void * (the corresponding itype values are PM_TYPE_64, PM_TYPE_U64,
56 PM_TYPE_DOUBLE, PM_TYPE_STRING, PM_TYPE_AGGREGATE and PM_TYPE_EVENT re‐
57 spectively). If valfmt specifies PM_VAL_PTR, then the value will be
58 extracted from the associated pmValueBlock structure, and the
59 __int32_t, __uint32_t and float options (itype being PM_TYPE_32,
60 PM_TYPE_U32 and PM_TYPE_FLOAT respectively) are not allowed, as
61 PM_VAL_INSITU is the appropriate encoding for these.
62
63 The following table defines the various possibilities for the type con‐
64 version -- the input type (itype) is shown vertically, and the output
65 type (otype) is shown horizontally. Y means the conversion is always
66 acceptable, N means the conversion can never be performed (the function
67 returns PM_ERR_CONV), P means the conversion may lose accuracy (but no
68 error status is returned), T means the result may be subject to high-
69 order truncation (in which case the function returns PM_ERR_TRUNC) and
70 S means the conversion may be impossible due to the sign of the input
71 value (in which case the function returns PM_ERR_SIGN). If an error
72 occurs, the value represented by oval will be zero (or NULL).
73
74 Note that although some of the conversions involving the types
75 PM_TYPE_STRING and PM_TYPE_AGGREGATE are indeed possible, but are
76 marked N - the rationale is that pmExtractValue should not be attempt‐
77 ing to duplicate functionality already available in the C library via
78 sscanf(3) and sprintf(3).
79
80 No conversion involving the type PM_TYPE_EVENT is supported.
81
82 | 32 | U32 | 64 | U64 | FLOAT | DBLE | STRNG | AGGR | EVENT
83 ======|=====|=======|=====|=======|=======|======|=======|======|=======
84 32 | Y | S | Y | S | P | P | N | N | N
85 U32 | T | Y | Y | Y | P | P | N | N | N
86 64 | T | T,S | Y | S | P | P | N | N | N
87 U64 | T | T | T | Y | P | P | N | N | N
88 FLOAT | P,T | P,T,S | P,T | P,T,S | Y | Y | N | N | N
89 DBLE | P,T | P,T,S | P,T | P,T,S | P | Y | N | N | N
90 STRNG | N | N | N | N | N | N | Y | N | N
91 AGGR | N | N | N | N | N | N | N | Y | N
92 EVENT | N | N | N | N | N | N | N | N | N
93
94 In the cases where multiple conversion errors could occur, the first
95 encountered error will be notified, and the order of checking is not
96 defined.
97
98 If the output conversion is to one of the pointer types, i.e. otype is
99 PM_TYPE_STRING or PM_TYPE_AGGREGATE, then the value buffer will have
100 been allocated by pmExtractValue(3) using malloc(3), and it is the
101 caller's responsibility to free the space when it is no longer re‐
102 quired.
103
104 Although this function appears rather complex, it has been constructed
105 to assist the development of performance tools that wish to convert
106 values, whose type is only known via the type field in a pmDesc struc‐
107 ture, into a canonical type for local processing.
108
110 PMAPI(3), pmAtomStr(3), pmConvScale(3), pmFetch(3), pmLookupDesc(3),
111 pmPrintValue(3), pmTypeStr(3), pmUnitsStr(3) and pmUnpackEven‐
112 tRecords(3).
113
115 PM_ERR_CONV
116
117 Impossible conversion, marked by N in above table
118
119 PM_ERR_TRUNC
120
121 High-order truncation occurred
122
123 PM_ERR_SIGN
124
125 Conversion of negative value to unsigned type attempted
126
127
128
129Performance Co-Pilot PCP PMEXTRACTVALUE(3)