On 12/05, Borislav Petkov wrote: > > On Thu, Dec 05, 2013 at 06:23:55PM +0100, Oleg Nesterov wrote: > > This is almost off-topic, but I am wondering if (in the long term) we > > can avoid this "insert the bp into every mm" altogether. > > > > Instead, uprobe_write_opcode() should only unmap this page and set > > Ok, sorry if I'm completely off base here but have you guys tried > unmapping the page from all other VMs,
This is what I meant, but we can't simply clear this pte, > and causing all > the VMs to refault why? it would be better to install the page on demand. > patching you'd probably need to cause the #PF handler to "loop" until > patching is complete though. We can't do this, but I do not think we need to block #PF handler. However, somehow the #PF handler should know that it should install the patched page owned by uprobes. That is why I talked about SWP_UPROBE_ENTRY (or something similar) But again, in any case this is not trivial. And perhaps I misundestood you... If you actually want to cause all the VMs to refault, then why we can't unmap + refault every mm like the patch I sent does? Just in case, note that we can't share the same page anyway without more complications. Oleg. -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
