Friday July 22, 2005
What's New in Solaris Express 7/05 (Nevada Build 17)
Today marks the release of Solaris Express 07/05 (or Nevada Build 17). You can obtain a free download.
We've got lots of cool new features on tap. Here's a breakdown. As
an experiment, I've started annotating some of the new features with links
to the corresponding RFE in the bug database.
Notable New Features in Solaris "Nevada", Build 17 (07/2005)
Desktop Support
- The AGPgart driver is integrated, providing greater graphics compatibility for Xorg. This feature improves support for memoryless graphics devices or devices which need to use system memory for video buffers (like Intel 810/830/855). Additionally, video drivers will be able to use AGPgart for 2D and 3D graphics acceleration.
- MESA support on x86 systems brings OpenGL support to a wide variety of Graphics hardware. Yay!
- Mozilla security advisories through MFSA 2005-42 have been fixed.
- dtlogin, dtconfig(1) and other CDE services have been converted to SMF:
svc:/application/graphical-login/dtlogin (dtlogin daemon) svc:/application/dtprintinfo (dtprintinfo service) svc:/network/dtspc (CDE Subprocess Control) svc:/network/rpc/calendar-manager (Calendar Manager) svc:/network/rpc/ttdbserver (ToolTalk)
Security
- Various cryptographic operations are faster. rc4 is faster on SPARC (more than 2x) [4925453]. SHA1 and MD5 should also be slightly faster [6264344].
- The sha2 kernel provider has been added, providing support for SHA256, SHA384, SHA512 [6181926].
Observability
- The NFS server now sets paths on its vnodes, meaning that you'll be more easily able to observe pathnames when using the DTrace I/O provider. [6175304]
- prtconf picks up a handy -b option for displaying the "product name" of a system, when available.
- dtrace now provides the fds[] array, a new stable DTrace variable which is defined as an array of fileinfo_t's indexed by integer. [6267695] Mike describes all of the details in his blog.
- new ls -S option to sort files by file size. [6262450]
Device Driver Interfaces
- The "Advanced DDI Interrupt Interface" was added (actually it's in 6/2005
but I missed it). Anish has
all of the details on the new APIs, but in short, a new DDI inteface family is provided
which has the following features:
- Supports new interrupt types like Message Signaled Interrupts (MSIs)
- Supports new interfaces for priority management, get/set interrupt capability, get interrupt pending info, set/clear interrupt mask
- Supports new I/O bus technologies like PCI-Express
- Introduces a generic framework that can support other new (and unknown) interrupt types where possible
- Supports multiple interrupts by a single device/function
- Maintains source and binary compatibility for existing DDI compliant leaf drivers
- Provides handle-based interrupt interfaces, which are more DDI compliant
Miscellaneous
- MPxIO interoperability with certain devices (including Engenio 2882,4884,5884) has been improved. [6254876]
- lex now outputs code less hostile to modern compilers [6267482]
- i.manifest and r.manifest are provided to ease the addition and removal of SMF manifests in packaging. [6209178]
- Lots of bugfixing and polishing around the "New Boot" architecture on x86. If you haven't upgraded to snv_15+, it's time!
Technorati Tag: Solaris
(2005-07-22 15:06:03.0) Permalink Comments [15]
Trackback: http://blogs.sun.com/dp/entry/what_s_new_in_solaris7


Posted by 192.65.17.24 on July 22, 2005 at 04:11 PM PDT #
Posted by Pete on July 22, 2005 at 08:25 PM PDT #
Posted by Peter W. Osel on July 22, 2005 at 11:03 PM PDT #
Posted by Matt Ingenthron on July 25, 2005 at 10:47 AM PDT #
Posted by Bill Webb on July 27, 2005 at 07:03 AM PDT #
Posted by Dan Price on July 27, 2005 at 12:23 PM PDT #
Posted by anonymous on July 28, 2005 at 11:10 PM PDT #
Posted by 192.18.42.11 on August 03, 2005 at 09:49 AM PDT #
Posted by Dan Price on August 04, 2005 at 04:22 PM PDT #
Posted by anonymous on August 05, 2005 at 11:35 PM PDT #
Posted by Jan Libo on August 13, 2005 at 08:26 AM PDT #
That's a bit insane IMO. Changing arbitrary code in your kernel is not a good idea unless you actually know what it is you're doing. Remember that you can't rely on this for anything but the exact bits the poster was using, as the compiled bits change all the time. Rebooting should clear the problem.
Posted by Dan Price on August 13, 2005 at 03:15 PM PDT #
Posted by Anonymous on October 25, 2005 at 09:51 PM PDT #
Posted by Douglas Atique on December 13, 2005 at 11:15 AM PST #
Posted by Douglas Atique on December 13, 2005 at 11:17 AM PST #