From guidoj2269 at gmail.com Mon Jun 1 20:58:39 2015 From: guidoj2269 at gmail.com (Guido) Date: Mon, 01 Jun 2015 20:58:39 +0200 Subject: [FreeVMS] x86_64 ia32 emulation In-Reply-To: <20150531203240.GA24151@sabre-wulf.nvg.ntnu.no> References: <5568E03C.4040808@gmail.com> <20150530131843.GL18627@sabre-wulf.nvg.ntnu.no> <556B1EBD.1040807@gmail.com> <20150531203240.GA24151@sabre-wulf.nvg.ntnu.no> Message-ID: <556CAB5F.7020604@gmail.com> On 05/31/2015 10:32 PM, Roar Thronæs wrote: >> I am trying to strip the code down to what is really needed and to add >> some automated tests to identify and fix these problems. The tests can >> also be used for regression testing. Anyway, this is quite lot of work, >> so do not expect new features any time soon. > Yes, I think the IA32 emulation may be stripped away for now. > Any other parts? > Mostly code for unsupported hardware or functionality that was not configured for the imported linux kernel. Several parts were removed previously and only some unused data structures or left over code will be removed. Guido From guidoj2269 at gmail.com Fri Jun 12 17:59:15 2015 From: guidoj2269 at gmail.com (Guido) Date: Fri, 12 Jun 2015 17:59:15 +0200 Subject: [FreeVMS] ioc$search Message-ID: <557B01D3.7020507@gmail.com> Hi all, The ioc$search function from iosubpagd.c should implement what is described in paragraphs 21.4.1 bullet 4 and 21.5.2.1 bullet 8, 9 & 10 from the VMS I&DS 5.2, right? It seems to try to extract the node name from the device name by searching for a '$', but the appropriate delimiter is '::'. If the node name delimiter is not found this will result in a null pointer. I do not understand the do-while loop searching the I/O database. I think the idea is that if a node name was found, it should match in the I/O database and that the device name and number must always match. But I believe the code does not quite do that. I intend to correct this piece of code, but I would like some confirmation that my analysis is correct. Regards, Guido From roart at nvg.ntnu.no Sat Jun 13 09:39:44 2015 From: roart at nvg.ntnu.no (Roar =?iso-8859-1?Q?Thron=C3=A6s?=) Date: Sat, 13 Jun 2015 09:39:44 +0200 Subject: [FreeVMS] ioc$search In-Reply-To: <557B01D3.7020507@gmail.com> References: <557B01D3.7020507@gmail.com> Message-ID: <20150613073944.GB24151@sabre-wulf.nvg.ntnu.no> On Fri, Jun 12, 2015 at 05:59:15PM +0200, Guido wrote: > > The ioc$search function from iosubpagd.c should implement what is > described in paragraphs 21.4.1 bullet 4 and 21.5.2.1 bullet 8, 9 & 10 > from the VMS I&DS 5.2, right? > > It seems to try to extract the node name from the device name by > searching for a '$', but the appropriate delimiter is '::'. If the node > name delimiter is not found this will result in a null pointer. > > I do not understand the do-while loop searching the I/O database. I > think the idea is that if a node name was found, it should match in the > I/O database and that the device name and number must always match. But > I believe the code does not quite do that. A lot of the code is temporary and workarounds, or only partially implemented. See http://h71000.www7.hp.com/doc/84final/4477/4477pro_010.html http://h71000.www7.hp.com/doc/731final/6489/6489pro_011.html The node$device is cluster device naming. http://h71000.www7.hp.com/doc/731final/6489/6489pro_007.html The node::disk is DECnet/OSI. -- Regards, Roar Thronæs From roart at nvg.ntnu.no Sat Jun 13 09:39:44 2015 From: roart at nvg.ntnu.no (Roar =?iso-8859-1?Q?Thron=C3=A6s?=) Date: Sat, 13 Jun 2015 09:39:44 +0200 Subject: [FreeVMS] ioc$search In-Reply-To: <557B01D3.7020507@gmail.com> References: <557B01D3.7020507@gmail.com> Message-ID: <20150613073944.GB24151@sabre-wulf.nvg.ntnu.no> On Fri, Jun 12, 2015 at 05:59:15PM +0200, Guido wrote: > > The ioc$search function from iosubpagd.c should implement what is > described in paragraphs 21.4.1 bullet 4 and 21.5.2.1 bullet 8, 9 & 10 > from the VMS I&DS 5.2, right? > > It seems to try to extract the node name from the device name by > searching for a '$', but the appropriate delimiter is '::'. If the node > name delimiter is not found this will result in a null pointer. > > I do not understand the do-while loop searching the I/O database. I > think the idea is that if a node name was found, it should match in the > I/O database and that the device name and number must always match. But > I believe the code does not quite do that. A lot of the code is temporary and workarounds, or only partially implemented. See http://h71000.www7.hp.com/doc/84final/4477/4477pro_010.html http://h71000.www7.hp.com/doc/731final/6489/6489pro_011.html The node$device is cluster device naming. http://h71000.www7.hp.com/doc/731final/6489/6489pro_007.html The node::disk is DECnet/OSI. -- Regards, Roar Thronæs From joel.bertrand at systella.fr Mon Jun 15 22:13:57 2015 From: joel.bertrand at systella.fr (=?UTF-8?Q?BERTRAND_Jo=c3=abl?=) Date: Mon, 15 Jun 2015 22:13:57 +0200 Subject: [FreeVMS] ioc$search In-Reply-To: <20150613073944.GB24151@sabre-wulf.nvg.ntnu.no> References: <557B01D3.7020507@gmail.com> <20150613073944.GB24151@sabre-wulf.nvg.ntnu.no> Message-ID: <557F3205.1070909@systella.fr> Roar, You can only send your messages to freevms at systella.fr. Your messages are always reveived twice, to freevms at rayleigh.systella.fr and freevms at systella.fr. Regards, JKB