|
I"m getting an error trying to run the DIFF MODULE on the Y-disk, and I could use some help figuring out what's wrong.? The error is DMSSMN109S VIRTUAL STORAGE CAPACITY EXCEEDED, but it doesn't matter how big I make the machine, it's always the same.? The error is coming from an OS GETMAIN SVC (SVC 10) call in what I think is the C runtime library.? It appears to be in the early setup of?@@CRT0, which reads: L0? ? ? ?B? ? ?12(,15) ? ? ? ? ?DC? ? C'@@CRT0' L12? ? ? STM? ?14,12,12(13) ? ? ? ? ?LR? ? 10,15 ? ? ? ? ?LR? ? 11,1 ? ? ? ? ?LR? ? 8,0 ? ? ? ? ?LR? ? 9,13 ? ? ? ? ?BAL? ?R1,L32 ? ? ? ? ?DC? ? A(X'1F534') L32? ? ? L? ? ?0,0(,1) ? ? ? ? ?SVC? ?10 That 1F534 pointer points to a bunch of words of all zeroes.? This whole thing looks screwy?to me.? Any ideas what's up? Ross |
开云体育1F534 should be the number of bytes it’s trying to allocate. The only thing screwy to me is the first instruction because I don’t get the padding around the entry symbol name plus the B to equal 12 but the assembler is probably aligning the next instruction by a rule I don’t remember.?I don’t have any idea how CMS emulates GETMAIN storage pools to know if there’s a big enough storage pool allocated. ?I’d look in the CMS manuals for help because you probably have to tailor the pool before running the command using it. On Mar 12, 2024, at 6:02?PM, Ross Patterson <ross.patterson@...> wrote:
|
开云体育Ross, I will say this is part of VM/380 so perhaps it only works in VM/380 and needs a machine in 31-bit mode to run. Why not use compare? Dave ? From: [email protected] <[email protected]> On Behalf Of Ross Patterson
Sent: Tuesday, March 12, 2024 10:02 PM To: [email protected] Subject: [h390-vm] DMSSMN109S in @@CRT0 running DIFF ? I"m getting an error trying to run the DIFF MODULE on the Y-disk, and I could use some help figuring out what's wrong.? The error is DMSSMN109S VIRTUAL STORAGE CAPACITY EXCEEDED, but it doesn't matter how big I make the machine, it's always the same.? The error is coming from an OS GETMAIN SVC (SVC 10) call in what I think is the C runtime library.? It appears to be in the early setup of?@@CRT0, which reads: ? L0? ? ? ?B? ? ?12(,15) ? ? ? ? ?DC? ? C'@@CRT0' L12? ? ? STM? ?14,12,12(13) ? ? ? ? ?LR? ? 10,15 ? ? ? ? ?LR? ? 11,1 ? ? ? ? ?LR? ? 8,0 ? ? ? ? ?LR? ? 9,13 ? ? ? ? ?BAL? ?R1,L32 ? ? ? ? ?DC? ? A(X'1F534') L32? ? ? L? ? ?0,0(,1) ? ? ? ? ?SVC? ?10 ? That 1F534 pointer points to a bunch of words of all zeroes.? This whole thing looks screwy?to me.? Any ideas what's up? ? Ross ? |
|
Paul Edwards says the code looks fine to him. Joe On Wed, Mar 13, 2024 at 3:04?AM Dave Wade <dave.g4ugm@...> wrote:
|
开云体育Joe, Well it worked for me on a couple of small files. Possibly ?just running out of memory. These ports of UNIX utilities all seem half baked to me. They do what Paul needs to build GCC but when you let them loose in other situations they may not work as expected. Dave ? From: [email protected] <[email protected]> On Behalf Of Joe Monk
Sent: Wednesday, March 13, 2024 12:00 PM To: [email protected] Subject: Re: [h390-vm] DMSSMN109S in @@CRT0 running DIFF ? Paul Edwards says the code looks fine to him. ? Joe ? On Wed, Mar 13, 2024 at 3:04?AM Dave Wade <dave.g4ugm@...> wrote:
|
|
On Wed, Mar 13, 2024 at 01:04 AM, Dave Wade wrote:
I will say this is part of VM/380 so perhaps it only works in VM/380 and needs a machine in 31-bit mode to runI thought so too but, actually, it works on VM/370 CE provided the virtual machine has the "right" amount of storage. I tried on CMSUSER with storage = 15M and the error message disappears (my previous failed attempt had CMSUSER with storage = 8M). I only checked on small files. Cheers, Rene FERLAND, Montreal |
|
On Tue, Mar 12, 2024 at 07:42 PM, Matthew T. Kromer wrote:
I don’t have any idea how CMS emulates GETMAIN storage pools to know if there’s a big enough storage pool allocated. ?I’d look in the CMS manuals for help because you probably have to tailor the pool before running the command using it.CMS GETMAIN starts after the end of the loaded module and works its way to high memory stopping at the loader tables. How much it grabs is a function of the actual request, and it returns an error if there isn't enough memory to satisfy the request. I don't remember if that is an abend, or just an error return. ?... Mark S. |
|
Hello, The code looks to me, but I used DMSFREE to obtain memory (most of the time). Bertram Moshier WB8ERT On Wed, Mar 13, 2024, 07:22 Dave Wade <dave.g4ugm@...> wrote:
|
|
On Wed, Mar 13, 2024 at 08:22 Dave Wade <dave.g4ugm@...> wrote:
Where did DIFF MODULE Y come from?? I can't find any source for it, it doesn't use the C libraries that VM/370 CE includes, and in fact it's statically linked with some library that doesn't ?match either of them.? I disassembled malloc(), which is driving the ABEND, and it doesn't resemble the assembler code in either PDPCLIB or GCCLIB/CMS. I'm going to switch to XCOMPARE for the case in working on, but I'm going to keep trying to figure this stuff out, because I'd love to see CMS get some of the more useful GNU/etc. tools. Ross? |
开云体育Ross, Leave that with me. I have found the source in the past. Dave ? From: [email protected] <[email protected]> On Behalf Of Ross Patterson
Sent: Thursday, March 14, 2024 2:29 AM To: [email protected] Subject: Re: [h390-vm] DMSSMN109S in @@CRT0 running DIFF ? On Wed, Mar 13, 2024 at 08:22 Dave Wade <dave.g4ugm@...> wrote:
Where did DIFF MODULE Y come from?? I can't find any source for it, it doesn't use the C libraries that VM/370 CE includes, and in fact it's statically linked with some library that doesn't ?match either of them.? I disassembled malloc(), which is driving the ABEND, and it doesn't resemble the assembler code in either PDPCLIB or GCCLIB/CMS. ? I'm going to switch to XCOMPARE for the case in working on, but I'm going to keep trying to figure this stuff out, because I'd love to see CMS get some of the more useful GNU/etc. tools. ? Ross? |
|
On Thu, Mar 14, 2024 at 10:28 AM, Ross Patterson wrote:
Where did DIFF MODULE Y come from? I can't find any source for it, itHi Ross. I'm not 100% sure I was involved in getting a DIFF executable onto a Y drive. I suspect I gave Robert O'Hara something at the time and he probably did that. I do remember one thing though. There was a problem with the build procedure (not the code) of producing GCCCMS (which I built) and I was attempting to allocate (I think 63 MiB) for an S/380 target when what was needed was the S/370 target which didn't do a single large malloc. And I wasn't alerted to the problem because the 63 MiB was being truncated at runtime to 15 MiB and succeeding in whatever environment I was using. But that issue (reported by Peppe) was fixed many years ago. But I'm not sure anyone has rebuilt anything since then and this was presumably pre-fix. Regardless, assuming it is indeed the version that I ship, the latest source is in custom.zip at - look for diffutils*. If you have a decent CMS environment (unlikely), then you should be able to unzip that and cd to src and type in "allcms". If you don't have a decent environment (likely), then you need to do the equivalent of everything in there. BFN. Paul. |
|
If the module is assuming vm/380 rather than vm/370 then in vmce or vm sixpack ,
and if you are running the SDL Hyperion version of Hercules, then in your Hercules .conf file, set MAINSIZE to say 512 rather than 16 (it may work with less but I am not sure) and add the following statement after the ARCHMODE line: FACILITY ENABLE herc_370_extension This should make VM ipl as VM/380 instead of VM/370, and enable use of the modified GETMAIN in VM/380. Please note I have no special knowledge of @@CRT0, this suggestion is based purely on user experience of VM/380 |
|
On Thu, Mar 14, 2024 at 10:19 PM, <eddy_balem@...> wrote:
If the module is assuming vm/380 rather than vm/370 then in vmce or vm sixpackThe split DAT has been added? Regardless, the only 31-bit CMS executable I ship is GCCCMS. All others are standard S/370 because they don't need lots of memory. BFN. Paul. |
|
On Thu, Mar 14, 2024 at 9:45?AM Paul Edwards <mutazilah@...> wrote: On Thu, Mar 14, 2024 at 10:28 AM, Ross Patterson wrote: Quite?by accident, I found that the DIFF MODULE dated 2009-06-06 on the VM/370 CE 1.2 Y disk comes from the 591 disk on the GCCCMS user, which is described in GCCCMS MEMO on its 191 disk as "GCC compiler version 3.2.3 MVS 7.0 from 2009".? No source so far, but baby steps are OK too. :-) ? Regardless, assuming it is indeed the version that I ship, the Yup, I see that.? It'll give me something to play with in my copious spare time :-) Thanks, Ross |
René Ferland
Mark A. Stevens