On 01/12/2022 12:12, Ayan Kumar Halder wrote:
On 01/12/2022 10:26, Julien Grall wrote:
Hi Ayan,
Hi Julien,
I have a question.
On 01/12/2022 10:03, Ayan Kumar Halder wrote:
It is possible for a pointer to represent physical memory of the same
size.
In other words, a 32 bit pointer can represent 32 bit addressable
physical
memory.
Thus, issue a compilation failure only when the count of physical
address bits
is greater than BITS_PER_LONG (ie count of bits in void*).
I am having difficult to understand how this description is related to
the BUILD_BUG_ON(). AFAIU, it is used to check that xenheap_bits can
be used in shift.
If the unsigned long is 32-bit, then a shift of 32 could be undefined.
Looking at the current use, the shift are used with "xenheap_bits -
PAGE_SHIFT". So as long as PAGE_SHIFT is not 0, you would be fine.
Ack
Signed-off-by: Ayan Kumar Halder <[email protected]>
---
Currently this change will not have any impact on the existing
architectures.
The following table illustrates PADDR_BITS vs BITS_PER_LONG of
different archs
------------------------------------------------
| Arch | PADDR_BITS | BITS_PER_LONG |
------------------------------------------------
| Arm_64 | 48 | 64 |
| Arm_32 | 40 | 32 |
| RISCV_64 | Don't know | 64 |
| x86 | 52 | 64 |
-------------------------------------------------
The Arm_32 line is a bit confusing because one would wonder why we
haven't seen this issue yet. So I think you want to clarify that the
code path is not used by Arm32.
Do you want this clarification and the above/below explanation to be a
part of the commit message ?
No I don't think this will be relevant in the final commit message.
Cheers,
--
Julien Grall
