Re: [pypy-dev] PPC stacklet switching code
Thanks Armin! This helps me a lot, I will try this out Regards, Vaibhav -Original Message- From: armin.r...@gmail.com [mailto:armin.r...@gmail.com] On Behalf Of Armin Rigo Sent: Wednesday, October 14, 2015 7:30 PM To: Vaibhav Sood Cc: pypy-dev@python.org Subject: Re: [pypy-dev] PPC stacklet switching code Hi Vaibhav, On Wed, Oct 14, 2015 at 11:26 AM, Vaibhav Sood wrote: > code to return back to the calling function slp_switch. However in > between the calling function’s stack content is overwritten by the > g_restore_state code (it memcpy’s the stored heap memory data to the > stack) causing the saved LR value to be destroyed and thus returning > from the g_restore_state function (using blr) causes a segfault I would decrement sp around the two calls to the C functions, g_save_state and g_restore_state, to give them their own temporary places to write to without interference. This is similar to how the x86 version gives "rsp" as argument, but then uses CALL, which pushes the return address just below rsp, i.e. out of scope of the saving/restoring. Example: ... mr r3, sp# the 1st argument is the value of sp without the extra space stdu sp, -64(sp) # decrement sp again; 64 should be enough ...call g_save_state... addi sp, sp, 64 # restore sp ... A bientôt, Armin. DISCLAIMER == This e-mail may contain privileged and confidential information which is the property of Persistent Systems Ltd. It is intended only for the use of the individual or entity to which it is addressed. If you are not the intended recipient, you are not authorized to read, retain, copy, print, distribute or use this message. If you have received this communication in error, please notify the sender and delete all copies of this message. Persistent Systems Ltd. does not accept any liability for virus infected mails. ___ pypy-dev mailing list pypy-dev@python.org https://mail.python.org/mailman/listinfo/pypy-dev
Re: [pypy-dev] PPC stacklet switching code
Sure, I have been referring to the ARM file (since it’s the one similar to PPC), I will try and take care of including the necessary header/footer and stack adjusting code, thanks for mentioning this issue! Regards, Vaibhav -Original Message- From: armin.r...@gmail.com [mailto:armin.r...@gmail.com] On Behalf Of Armin Rigo Sent: Wednesday, October 14, 2015 7:38 PM To: Vaibhav Sood Cc: pypy-dev@python.org Subject: Re: [pypy-dev] PPC stacklet switching code Re-hi, Note another issue: the slp_switch() function, from gcc's point of view, does not call any other function. This means gcc will not produce the needed header/footer code. Look at switch_arm_gcc.h for the lengths we need to go through to make sure to save/restore everything and realign the stack. A bientôt, Armin. DISCLAIMER == This e-mail may contain privileged and confidential information which is the property of Persistent Systems Ltd. It is intended only for the use of the individual or entity to which it is addressed. If you are not the intended recipient, you are not authorized to read, retain, copy, print, distribute or use this message. If you have received this communication in error, please notify the sender and delete all copies of this message. Persistent Systems Ltd. does not accept any liability for virus infected mails. ___ pypy-dev mailing list pypy-dev@python.org https://mail.python.org/mailman/listinfo/pypy-dev
Re: [pypy-dev] PPC stacklet switching code
Re-hi, Note another issue: the slp_switch() function, from gcc's point of view, does not call any other function. This means gcc will not produce the needed header/footer code. Look at switch_arm_gcc.h for the lengths we need to go through to make sure to save/restore everything and realign the stack. A bientôt, Armin. ___ pypy-dev mailing list pypy-dev@python.org https://mail.python.org/mailman/listinfo/pypy-dev
Re: [pypy-dev] PPC stacklet switching code
Hi Vaibhav, On Wed, Oct 14, 2015 at 11:26 AM, Vaibhav Sood wrote: > code to return back to the calling function slp_switch. However in between > the calling function’s stack content is overwritten by the g_restore_state > code (it memcpy’s the stored heap memory data to the stack) causing the > saved LR value to be destroyed and thus returning from the g_restore_state > function (using blr) causes a segfault I would decrement sp around the two calls to the C functions, g_save_state and g_restore_state, to give them their own temporary places to write to without interference. This is similar to how the x86 version gives "rsp" as argument, but then uses CALL, which pushes the return address just below rsp, i.e. out of scope of the saving/restoring. Example: ... mr r3, sp# the 1st argument is the value of sp without the extra space stdu sp, -64(sp) # decrement sp again; 64 should be enough ...call g_save_state... addi sp, sp, 64 # restore sp ... A bientôt, Armin. ___ pypy-dev mailing list pypy-dev@python.org https://mail.python.org/mailman/listinfo/pypy-dev