[Qemu-devel] [PATCH 00/13] hw/block/pflash_cfi02: Clean-up and fixes

Philippe Mathieu-Daudé posted 13 patches 4 years, 11 months ago
Test docker-mingw@fedora passed
Test checkpatch passed
Test docker-clang@ubuntu passed
Test asan passed
Patches applied successfully (tree, apply log)
git fetch https://github.com/patchew-project/qemu tags/patchew/20190505221544.31568-1-philmd@redhat.com
Maintainers: Paolo Bonzini <pbonzini@redhat.com>, Max Reitz <mreitz@redhat.com>, Thomas Huth <thuth@redhat.com>, Kevin Wolf <kwolf@redhat.com>, Laurent Vivier <lvivier@redhat.com>
hw/block/pflash_cfi02.c   | 234 +++++++++++++++++---------------------
tests/Makefile.include    |   2 +
tests/pflash-cfi02-test.c | 232 +++++++++++++++++++++++++++++++++++++
3 files changed, 339 insertions(+), 129 deletions(-)
create mode 100644 tests/pflash-cfi02-test.c
[Qemu-devel] [PATCH 00/13] hw/block/pflash_cfi02: Clean-up and fixes
Posted by Philippe Mathieu-Daudé 4 years, 11 months ago
Hi,

While reviewing Stephen Checkoway's v4 "Implement missing AMD
pflash functionality" [*] I found it hard (for me) to digest,
so I took step by step notes. This series is the result of
those notes.
Regarding Stephen's series, this series only contains the
generic code movement and trivial cleanup. The other patches
are rather dense and I need more time to study the specs.

Stephen: If you take out the patch #2 ("Use the GLib API"),
you can rebase your series on top of this.
I'd appreciate if you can adapt your tests to use the GLib
functions, else I plan to do it later.

Regards,

Phil.

[*] https://lists.gnu.org/archive/html/qemu-devel/2019-04/msg04595.html

Philippe Mathieu-Daudé (10):
  tests/pflash-cfi02: Use the GLib API
  tests/pflash-cfi02: Use IEC binary prefixes for size constants
  hw/block/pflash_cfi02: Fix debug format string
  hw/block/pflash_cfi02: Add an enum to define the write cycles
  hw/block/pflash_cfi02: Add helpers to manipulate the status bits
  hw/block/pflash_cfi02: Simplify a statement using fall through
  hw/block/pflash_cfi02: Use the ldst API in pflash_write()
  hw/block/pflash_cfi02: Use the ldst API in pflash_read()
  hw/block/pflash_cfi02: Extract the pflash_data_read() function
  hw/block/pflash_cfi02: Unify the MemoryRegionOps

Stephen Checkoway (3):
  tests/pflash-cfi02: Add test for supported CFI commands
  hw/block/pflash_cfi02: Fix command address comparison
  hw/block/pflash_cfi02: Use the chip erase time specified in the CFI
    table

 hw/block/pflash_cfi02.c   | 234 +++++++++++++++++---------------------
 tests/Makefile.include    |   2 +
 tests/pflash-cfi02-test.c | 232 +++++++++++++++++++++++++++++++++++++
 3 files changed, 339 insertions(+), 129 deletions(-)
 create mode 100644 tests/pflash-cfi02-test.c

-- 
2.20.1


Re: [Qemu-devel] [Qemu-block] [PATCH 00/13] hw/block/pflash_cfi02: Clean-up and fixes
Posted by John Snow 4 years, 10 months ago
I don't think this series ever made it upstream, and it's now well past
30 days, so I might encourage a resend when you can if this is still
important to pursue.

--js

On 5/5/19 6:15 PM, Philippe Mathieu-Daudé wrote:
> Hi,
> 
> While reviewing Stephen Checkoway's v4 "Implement missing AMD
> pflash functionality" [*] I found it hard (for me) to digest,
> so I took step by step notes. This series is the result of
> those notes.
> Regarding Stephen's series, this series only contains the
> generic code movement and trivial cleanup. The other patches
> are rather dense and I need more time to study the specs.
> 
> Stephen: If you take out the patch #2 ("Use the GLib API"),
> you can rebase your series on top of this.
> I'd appreciate if you can adapt your tests to use the GLib
> functions, else I plan to do it later.
> 
> Regards,
> 
> Phil.
> 
> [*] https://lists.gnu.org/archive/html/qemu-devel/2019-04/msg04595.html
> 
> Philippe Mathieu-Daudé (10):
>   tests/pflash-cfi02: Use the GLib API
>   tests/pflash-cfi02: Use IEC binary prefixes for size constants
>   hw/block/pflash_cfi02: Fix debug format string
>   hw/block/pflash_cfi02: Add an enum to define the write cycles
>   hw/block/pflash_cfi02: Add helpers to manipulate the status bits
>   hw/block/pflash_cfi02: Simplify a statement using fall through
>   hw/block/pflash_cfi02: Use the ldst API in pflash_write()
>   hw/block/pflash_cfi02: Use the ldst API in pflash_read()
>   hw/block/pflash_cfi02: Extract the pflash_data_read() function
>   hw/block/pflash_cfi02: Unify the MemoryRegionOps
> 
> Stephen Checkoway (3):
>   tests/pflash-cfi02: Add test for supported CFI commands
>   hw/block/pflash_cfi02: Fix command address comparison
>   hw/block/pflash_cfi02: Use the chip erase time specified in the CFI
>     table
> 
>  hw/block/pflash_cfi02.c   | 234 +++++++++++++++++---------------------
>  tests/Makefile.include    |   2 +
>  tests/pflash-cfi02-test.c | 232 +++++++++++++++++++++++++++++++++++++
>  3 files changed, 339 insertions(+), 129 deletions(-)
>  create mode 100644 tests/pflash-cfi02-test.c
> 

Re: [Qemu-devel] [Qemu-block] [PATCH 00/13] hw/block/pflash_cfi02: Clean-up and fixes
Posted by Philippe Mathieu-Daudé 4 years, 10 months ago
Hi John,

On 6/26/19 10:33 PM, John Snow wrote:
> I don't think this series ever made it upstream, and it's now well past
> 30 days, so I might encourage a resend when you can if this is still
> important to pursue.

I should have sent a 'ping' indeed.
I keep rebasing because I have it in my pflash-next queue, and I added
more patches from Stephen. I am still running tests, and it is a pain to
test things that have never been tested.
Anyway, my plan is to send a "current status of pflash-next" series.

Thanks for worrying :)

Phil.

> --js
> 
> On 5/5/19 6:15 PM, Philippe Mathieu-Daudé wrote:
>> Hi,
>>
>> While reviewing Stephen Checkoway's v4 "Implement missing AMD
>> pflash functionality" [*] I found it hard (for me) to digest,
>> so I took step by step notes. This series is the result of
>> those notes.
>> Regarding Stephen's series, this series only contains the
>> generic code movement and trivial cleanup. The other patches
>> are rather dense and I need more time to study the specs.
>>
>> Stephen: If you take out the patch #2 ("Use the GLib API"),
>> you can rebase your series on top of this.
>> I'd appreciate if you can adapt your tests to use the GLib
>> functions, else I plan to do it later.

PD: Oh, Stephen, if you read it, forget about this comment!


>> [*] https://lists.gnu.org/archive/html/qemu-devel/2019-04/msg04595.html
>>
>> Philippe Mathieu-Daudé (10):
>>   tests/pflash-cfi02: Use the GLib API
>>   tests/pflash-cfi02: Use IEC binary prefixes for size constants
>>   hw/block/pflash_cfi02: Fix debug format string
>>   hw/block/pflash_cfi02: Add an enum to define the write cycles
>>   hw/block/pflash_cfi02: Add helpers to manipulate the status bits
>>   hw/block/pflash_cfi02: Simplify a statement using fall through
>>   hw/block/pflash_cfi02: Use the ldst API in pflash_write()
>>   hw/block/pflash_cfi02: Use the ldst API in pflash_read()
>>   hw/block/pflash_cfi02: Extract the pflash_data_read() function
>>   hw/block/pflash_cfi02: Unify the MemoryRegionOps
>>
>> Stephen Checkoway (3):
>>   tests/pflash-cfi02: Add test for supported CFI commands
>>   hw/block/pflash_cfi02: Fix command address comparison
>>   hw/block/pflash_cfi02: Use the chip erase time specified in the CFI
>>     table
>>
>>  hw/block/pflash_cfi02.c   | 234 +++++++++++++++++---------------------
>>  tests/Makefile.include    |   2 +
>>  tests/pflash-cfi02-test.c | 232 +++++++++++++++++++++++++++++++++++++
>>  3 files changed, 339 insertions(+), 129 deletions(-)
>>  create mode 100644 tests/pflash-cfi02-test.c
>>

Re: [Qemu-devel] [Qemu-block] [PATCH 00/13] hw/block/pflash_cfi02: Clean-up and fixes
Posted by John Snow 4 years, 10 months ago

On 6/26/19 5:06 PM, Philippe Mathieu-Daudé wrote:
> Hi John,
> 
> On 6/26/19 10:33 PM, John Snow wrote:
>> I don't think this series ever made it upstream, and it's now well past
>> 30 days, so I might encourage a resend when you can if this is still
>> important to pursue.
> 
> I should have sent a 'ping' indeed.
> I keep rebasing because I have it in my pflash-next queue, and I added
> more patches from Stephen. I am still running tests, and it is a pain to
> test things that have never been tested.
> Anyway, my plan is to send a "current status of pflash-next" series.
> 
> Thanks for worrying :)
> 
> Phil.
> 

Great, thanks!