|

»

jikesrvm-researchers

[rvm-research] Methods address in compiledMethods array

View: New views

3 Messages — Rating Filter: Alert me

	[rvm-research] Methods address in compiledMethods array by Anthony Hocquet :: Rate this Message: Reply to Author \| View Threaded \| Show Only this Message Hi everyone, For my debugging purpose, I check methods address in CompiledMethods.compiledMethods[][]. I print the method address supposed to be inserted in setCompiledMethod() and print the method address read in snipObsoleteCompiledMethods(). I notice that methods inserted in setCompiledMethod() begins with 0x96/0x97 address mask whereas those read in snipObsoleteCompiledMethods() begins with 0x68 mask. Why? Shouldn't be the same methods addess? I modified the existing write barrier and I have a bug in snipObsoleteCompiledMethods : it reads a method with an address mask 0x96. It results in a hardware fault. It looks like the object referenced by this address is not a CompiledMethod instance. Best regards, Anthony Hocquet ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf _______________________________________________ Jikesrvm-researchers mailing list Jikesrvm-researchers@... https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers
	Re: [rvm-research] Methods address in compiledMethods array by Daniel Frampton :: Rate this Message: Reply to Author \| View Threaded \| Show Only this Message Objects can move. The production configuration is a generational collector, and additionally uses Immix which may defragment parts of the mature space. The 0x96/0x97 addresses look to be nursery addresses, while the lower addresses are probably in the mature space. In general, if you want to hold a reference to objects and you are not writing the garbage collector, you need to make sure those references are seen (and updated) by the collector. I am not sure what barrier changes you have made, but if you have made an error in the barrier so that it is not keeping track of references from the mature to the nursery space, then you could expect many sorts of errors including dangling pointers. Cheers, Daniel. On Thu, Sep 17, 2009 at 9:45 AM, Anthony Hocquet <anthony.hocquet@...> wrote: Hi everyone, For my debugging purpose, I check methods address in CompiledMethods.compiledMethods[][]. I print the method address supposed to be inserted in setCompiledMethod() and print the method address read in snipObsoleteCompiledMethods(). I notice that methods inserted in setCompiledMethod() begins with 0x96/0x97 address mask whereas those read in snipObsoleteCompiledMethods() begins with 0x68 mask. Why? Shouldn't be the same methods addess? I modified the existing write barrier and I have a bug in snipObsoleteCompiledMethods : it reads a method with an address mask 0x96. It results in a hardware fault. It looks like the object referenced by this address is not a CompiledMethod instance. Best regards, Anthony Hocquet ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf _______________________________________________ Jikesrvm-researchers mailing list Jikesrvm-researchers@... https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf _______________________________________________ Jikesrvm-researchers mailing list Jikesrvm-researchers@... https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers
	Re: [rvm-research] Methods address in compiledMethods array by Anthony Hocquet :: Rate this Message: Reply to Author \| View Threaded \| Show Only this Message Hi Daniel, You're right : objecs can move. I completely missed that fact. Indeed, there was a bug in my code. Some references from the mature to the nursery space were not insereted into the rembered sets. It results in a collection of living objets, which explains the segmentation faults I had. Now, thanks to you, I solved this issue. Thanks for your reply, it's really appreciated. Best regards, On Sep 17, 2009, at 2:26 AM, Daniel Frampton wrote: Objects can move. The production configuration is a generational collector, and additionally uses Immix which may defragment parts of the mature space. The 0x96/0x97 addresses look to be nursery addresses, while the lower addresses are probably in the mature space. In general, if you want to hold a reference to objects and you are not writing the garbage collector, you need to make sure those references are seen (and updated) by the collector. I am not sure what barrier changes you have made, but if you have made an error in the barrier so that it is not keeping track of references from the mature to the nursery space, then you could expect many sorts of errors including dangling pointers. Cheers, Daniel. On Thu, Sep 17, 2009 at 9:45 AM, Anthony Hocquet <anthony.hocquet@...> wrote: Hi everyone, For my debugging purpose, I check methods address in CompiledMethods.compiledMethods[][]. I print the method address supposed to be inserted in setCompiledMethod() and print the method address read in snipObsoleteCompiledMethods(). I notice that methods inserted in setCompiledMethod() begins with 0x96/0x97 address mask whereas those read in snipObsoleteCompiledMethods() begins with 0x68 mask. Why? Shouldn't be the same methods addess? I modified the existing write barrier and I have a bug in snipObsoleteCompiledMethods : it reads a method with an address mask 0x96. It results in a hardware fault. It looks like the object referenced by this address is not a CompiledMethod instance. Best regards, Anthony Hocquet ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf _______________________________________________ Jikesrvm-researchers mailing list Jikesrvm-researchers@... https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf_______________________________________________ Jikesrvm-researchers mailing list Jikesrvm-researchers@... https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers -- Anthony Hocquet ------------------------------------------------------------------------------ Come build with us! The BlackBerry® Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9-12, 2009. Register now! http://p.sf.net/sfu/devconf _______________________________________________ Jikesrvm-researchers mailing list Jikesrvm-researchers@... https://lists.sourceforge.net/lists/listinfo/jikesrvm-researchers