blob: c6b04937cee3a2e4cc79d48b5ae5c44fa1becad9 [file] [log] [blame]
Mauro Carvalho Chehabdc7a12b2019-04-14 15:51:10 -03001===============================================
Linus Torvalds1da177e2005-04-16 15:20:36 -07002Release notes for Linux Kernel VFP support code
Mauro Carvalho Chehabdc7a12b2019-04-14 15:51:10 -03003===============================================
Linus Torvalds1da177e2005-04-16 15:20:36 -07004
5Date: 20 May 2004
Mauro Carvalho Chehabdc7a12b2019-04-14 15:51:10 -03006
Linus Torvalds1da177e2005-04-16 15:20:36 -07007Author: Russell King
8
9This is the first release of the Linux Kernel VFP support code. It
10provides support for the exceptions bounced from VFP hardware found
11on ARM926EJ-S.
12
13This release has been validated against the SoftFloat-2b library by
14John R. Hauser using the TestFloat-2a test suite. Details of this
15library and test suite can be found at:
16
Randy Dunlap98766fb2005-11-21 21:32:31 -080017 http://www.jhauser.us/arithmetic/SoftFloat.html
Linus Torvalds1da177e2005-04-16 15:20:36 -070018
19The operations which have been tested with this package are:
20
21 - fdiv
22 - fsub
23 - fadd
24 - fmul
25 - fcmp
26 - fcmpe
27 - fcvtd
28 - fcvts
29 - fsito
30 - ftosi
31 - fsqrt
32
33All the above pass softfloat tests with the following exceptions:
34
35- fadd/fsub shows some differences in the handling of +0 / -0 results
36 when input operands differ in signs.
37- the handling of underflow exceptions is slightly different. If a
38 result underflows before rounding, but becomes a normalised number
39 after rounding, we do not signal an underflow exception.
40
41Other operations which have been tested by basic assembly-only tests
42are:
43
44 - fcpy
45 - fabs
46 - fneg
47 - ftoui
48 - ftosiz
49 - ftouiz
50
51The combination operations have not been tested:
52
53 - fmac
54 - fnmac
55 - fmsc
56 - fnmsc
57 - fnmul