I see! I needed to lengthen the list of packages to prevent yum from removing important packages. Thus, I needed to include the five packages
glibc glibc-common glibc-devel glibc-headers nscd in the list of packages. That worked. Thank you. > When I try that, yum warns me that 14 > dependent packages will be removed. > The packages include gcc stuff. > Should my yum line include all the > packages that /var/log/yum.log shows > were updated on December 12, 2011? > (The updates were done by the CentOS > updater that announces updates as > they become available.) > >> Hi Ron, >> >> I used yum's 'downgrade' command: >> >> $sudo yum downgrade glibc glibc-common >> >> This should downgrade to the next highest version. >> >> Mark >> >> >> On Mon, Dec 26, 2011 at 4:44 PM, <[email protected]> wrote: >> >>> How do you safely roll back glibc? I >>> would like to try your fix but also >>> want to avoid breaking any system >>> setups. I am using CentOS. Is there >>> a way to have Xilinx and Matlab >>> software use a rolled-back glibc but >>> allow other software to continue >>> using the current version? >>> >>> > I tried rolling back glibc and that did fix the XST error issue, but >>> other >>> > issues are now surfacing so i've opened a 'Webcase' with Xilinx and >>> will >>> > hopefully hear back from them soon. >>> > >>> > Mark >>> > >>> > >>> > On Sun, Dec 25, 2011 at 7:58 PM, John Ford <[email protected]> wrote: >>> > >>> >> > I need to check some logs, but I do >>> >> > think somebody did some of the >>> >> > CentOS updates that were announced >>> >> > "available" by the CentOS system >>> >> > updater. I don't see the reference >>> >> > to the glibc problem in recent >>> >> > CASPER email. >>> >> >>> >> May mail seems to have bounced that I sent. In any case, there is a >>> bug >>> >> in xst that double-frees some memory, triggering glibc to throw a >>> memory >>> >> exception and scream and die. We rolled back our glibc and it fixed >>> the >>> >> problem. Well, it masks the problem... >>> >> >>> >> Sorry about the lack of info! >>> >> >>> >> John >>> >> >>> >> >>> >> > >>> >> >>> A new problem is preventing me from >>> >> >>> compiling even simple models with >>> >> >>> bee_xps. For example, when I compile >>> >> >>> a model containing only the system >>> >> >>> generator, XSB Core Config, Xilinx >>> >> >>> constants and CASPER software >>> >> >>> registers, I receive the following >>> >> >>> messages from bee_xps. >>> >> >>> >>> >> >>> Detected Linux OS >>> >> >>> ############################# >>> >> >>> ## System Update ## >>> >> >>> ############################# >>> >> >>> ############################# >>> >> >>> ## Block objects creation ## >>> >> >>> ############################# >>> >> >>> ###################### >>> >> >>> ## Checking objects ## >>> >> >>> ###################### >>> >> >>> Running system generator ... >>> >> >>> Error using ==> gen_xps_files at 328 >>> >> >>> XSG generation failed: ? >>> >> >>> >>> >> >>> I am using Xilinx 11.1 and Matlab >>> >> >>> R2009a on CentOS Linux with the >>> >> >>> mlib_devel_10_1 library. I was not >>> >> >>> having this trouble until recently. >>> >> >>> Other people use my computer, but >>> >> >>> they know to avoid changing my >>> >> >>> installations. >>> >> >>> >>> >> >>> I attached a simple MDL file that >>> >> >>> produces the error messages shown >>> >> >>> above. What is causing gen_xps_files >>> >> >>> to have trouble at 328? >>> >> >>> >>> >> >>> >>> >> >> >>> >> >> Hi Ron. This sounds strangely like the glibc problem I mentioned >>> >> last >>> >> >> week. Did any messages from the OS come up in the window you >>> started >>> >> >> the >>> >> >> matlab session from? >>> >> >> >>> >> >> You might check to see if anyone updated your glibc lately. >>> >> >> >>> >> >> I'll try your mdl tomorrow and see what happens. I don't have >>> access >>> >> to >>> >> >> a >>> >> >> toolset right now. >>> >> >> >>> >> >> John >>> >> >> >>> >> >> >>> >> >> >>> >> > >>> >> > >>> >> >>> >> >>> >> >>> >> >>> > >>> >>> >>> >>> >> > > >
