What I see now when disassembling the two products (e.g., /kernel/drv/adm64/mpt_sas), there are subtle style differences in code arrangement:
mptsas_iport_attach()
mptsas_iport_attach: 55 pushq %rbp
mptsas_iport_attach+0x1: 48 8b ec movq %rsp,%rbp
mptsas_iport_attach+0x4: 48 83 ec 10 ** subq $0x10,%rsp
mptsas_iport_attach+0x8: 48 89 7d f8 movq %rdi,-0x8(%rbp)
mptsas_iport_attach+0xc: 48 89 75 f0 movq %rsi,-0x10(%rbp)
mptsas_iport_attach()
mptsas_iport_attach: 55 pushq %rbp
mptsas_iport_attach+0x1: 48 8b ec movq %rsp,%rbp
mptsas_iport_attach+0x4: 48 89 7c 24 f8 movq %rdi,-0x8(%rsp)
mptsas_iport_attach+0x9: 48 89 74 24 f0 movq %rsi,-0x10(%rsp)
mptsas_iport_attach+0xe: 48 83 ec 10 ** subq $0x10,%rsp
From this I conclude that the oi_151a9 /dev/ official product is
produced with a different Sun CC version than the one mentioned in
the illumos build instructions. I have
cc: Sun C 5.9 SunOS_i386 Patch 124868-10 2009/04/30
reconstructed from a vanilla 12.0 compiler and the patches
124868-10 126496-03 126498-02
as well as the 12.1u1 lint checker
lint: Sun C 5.10 SunOS_i386 Patch 142363-03 2009/12/03
constructed from vanilla 12.1u1 with patches
141858-03 142363-03
Which one is the compiler oi_141a9 /dev is compiled with ?
And, if different, is it approved to do The Right Thing ?
On 18/06/2015 11:38, Udo Grabowski (IMK) wrote:
Hi, I'm trying to reproduce the exact illumos kernel currenly used in /dev a9 (that is, a9 update 2 with the bash fixes from early this year, SUN CC compiled). I have the necessary compiler versions (reconstructing it was a fun archeological adventure into SolarisExpress paleolithicum...), and cloned this repository, which seems like the right one judging from the commit dates: <https://hg.openindiana.org/sustaining/oi_151a/illumos-gate/> illumos rev 52e13e00ba (bash fixes etc. are in sfw-gate/) After compiling (on exactly the same oi I want to reproduce) according to the instructions in <http://wiki.illumos.org/display/illumos/How+To+Build+illumos>, setting LANG and LC_ALL to C to get around the locale difference, fixing the jdk7 problem (#4719), setting __SUNC="", switching off DEBUG build, NIGHTLY_OPTIONS='-nClrpt', I was lucky to see that mpt and marvell88sx were exactly reproduced (checked md5sum): size 387160 2012-10-08 14:34 /kernel/drv/amd64/mpt* 95112 2012-10-08 14:34 /kernel/drv/amd64/marvell88sx* But a short inspection of the other drivers quickly told me that I must have somehow pulled the wrong gate (note the date difference, only the 2012 entries and the small mouse8042 are reproduced, not the new ones): ..... Where is the real gate used for the current kernel in oi 151a9 /dev ? Or is this another difference somewhere ?
-- Dr.Udo Grabowski Inst.f.Meteorology & Climate Research IMK-ASF-SAT http://www.imk-asf.kit.edu/english/sat.php KIT - Karlsruhe Institute of Technology http://www.kit.edu Postfach 3640,76021 Karlsruhe,Germany T:(+49)721 608-26026 F:-926026
smime.p7s
Description: S/MIME Cryptographic Signature
_______________________________________________ oi-dev mailing list [email protected] http://openindiana.org/mailman/listinfo/oi-dev
