Re: RFC: Adding non-PIC executable support to MIPS

[Richard Sandiford <rdsandiford@googlemail.com>]

Mark Mitchell <mark@codesourcery.com> writes:
> Richard Sandiford wrote:
>> Daniel Jacobowitz <dan@debian.org> writes:
>>> All comments welcome - Richard, especially from you.  How would you
>>> like to proceed?  I think the first step should be to get your other
>>> binutils/gcc patches merged, including MIPS16 PIC; I used those as a
>>> base.  But see a few of the notes for potential problems with those
>>> patches.
>> 
>> Yeah, Nick's approved most of the remaining binutils changes (thanks).
>> I haven't applied them yet because of the doubt over whether st_size
>> should be even or odd for ISA-encoded MIPS16 symbols.  I don't really
>> have an opinion, so I'll accept a maintainerly decision...
>
> [I'm not sure if this is a helpful suggestion or not, so feel free to 
> ignore it if it's not.]
>
> I would suggest that st_size be the actual size of the function, as it 
> lives in memory.  A test of it's start/end location is "could I stick a 
> random data byte there and have it affect the function".  For example, 
> for a Thumb function whose ISA address is "0x00000001", I would consider 
> for size purposes that it starts at "0x00000000", since altering that 
> byte at run-time would change the meaning of the function.

For the record, my reasoning when picking the odd st_size was similar,
but with the opposite outcome.  The point of using an ISA-encoded
st_value is that that's what most users want.  Most of them won't
even have code to say "is this a MIPS16 symbol?".

So if users are going to get into the habit of using MIPS st_values
without checking the "ISA bit", I thought it was more conservative to
base the end address on the unmodified st_value rather than the modified
one.  In other words, I thought it was more conservative to have
"st_value + st_size" be the end point of the function, rather than
"(st_value & ~1) + st_size".  This ensures that "st_value" and
"st_value + st_size - 1" are bytes in the function, rather than making
"st_value + st_size" be two bytes past the end of the function (and thus
making "st_value + st_size - 1" refer to something outside the function).

But like I say, I can see there are pros and cons both ways, so I don't
really have an opinion.  I'm happy to (and do) accpet Dan's decision.
And I guess the ARM experience shows that my concern isn't really an
issue in practice anyway.

Richard