Hi Thomas and List, Thomas Petazzoni <thomas.petazzoni <at> free-electrons.com> writes: > [...] > > I suspect that the argument "struct page *" is passed in %rdi. It first > dereferences it to get the page->flags value (which is at the beginning > of the page structure). Once this value is in %rax, it shifts it right > by 38 bits, which I suspect is the >> ZONETABLE_PGSHIFT in your source > code. Then, it accesses to the zone_table[] array : it access to the > %rax-th entry, in an array of elements with an 8 byte size (and > zone_table[] is an array of "struct zone *", so it's an array of 8 > bytes elements). In %rdx, we now have a pointer to a "struct zone *"... > which in fact happens to be NULL. Thanks for the explanation, now it makes sense to me. > > So, I would say that either your page->flags are broken, either your > zone_table[] array is incorrect. Yup, The flags field of the page was broken and the value was going beyond the available zone_table index. And hence the problem in dereferencing. > > > 2) for "RIP: 0010:[<ffffffff801212a7>] " what does > > "0010" mean? > > It's the segment selector. It selects through which segment we are > accessing the code. It is an offset in the Global Descriptor Table > (GDT). These are x86/x86_64 specific bits, I don't think you care about > this for your particular issue. Actually yes, I doesn't relate to my problem but anyways, thanks. > > > 3) Does "EFLAGS: 00010217" signify any error value, and > > if it does, how to interpret it? > > Take the Intel Developer manual, volume 1 (available at > http://www.intel.com/products/processor/manuals/), page 78. It > describes the contents of the EFLAGS register. Done. > > Sincerly, > > Thomas And sorry, Cannot put Thomas in CC as I am mailing from the Gmane.org web interface and it doesn't give me any option for that. Your help is highly appreciated. Shreyansh -- To unsubscribe from this list: send an email with "unsubscribe kernelnewbies" to ecartis@xxxxxxxxxxxx Please read the FAQ at http://kernelnewbies.org/FAQ
