Re: mainline build failure due to f1e4c916f97f ("drm/edid: add EDID block count and size helpers")

[Keisuke Nishimura]

On 2022/06/01 1:41, Linus Torvalds wrote:
> On Tue, May 31, 2022 at 1:04 AM Arnd Bergmann <arnd@xxxxxxxx> wrote:
> > As an experiment: what kind of results would we get when looking
> > for packed structures and unions that contain any of these:

> I don't think we have that. Not only because it would already cause
> breakage, but simply because the kinds of structures that people pack
> aren't generally the kind that contain these kinds of things.
>
> That said, you might have a struct that is packed, but that
> intentionally aligns parts of itself, so it *could* be valid.
>
> But it would probably not be a bad idea to check that packed
> structures/unions don't have atomic types or locks in them. I _think_
> we're all good, but who knows..


I am Julia's student at INRIA and I heard from her that there is an 
opportunity to use Coccinelle to find specific types in packed struct or
enum.

I found 13 definitions of packed structure that contains:
> - spinlock_t
> - atomic_t
> - dma_addr_t
> - phys_addr_t
> - size_t

> - struct mutex
> - struct device

- raw_spinlock_t

== Results ==
security/tomoyo/common.h: atomic_t in tomoyo_shared_acl_head
drivers/net/ethernet/chelsio/inline_crypto/chtls/chtls.h: spinlock_t in 
key_map
include/linux/ti-emif-sram.h: phys_addr_t in ti_emif_pm_data
drivers/scsi/wd719x.h: dma_addr_t in wd719x_scb
drivers/net/wireless/intel/ipw2x00/ipw2200.h: dma_addr_t in clx2_queue
drivers/infiniband/hw/irdma/osdep.h: dma_addr_t in irdma_dma_mem
drivers/infiniband/core/mad_priv.h: size_t in ib_mad_private
drivers/crypto/qat/qat_common/qat_asym_algs.c:
- dma_addr_t in qat_rsa_ctx
- dma_addr_t in qat_dh_ctx
drivers/atm/idt77252.h: dma_addr_t in idt77252_skb_prv
arch/s390/include/asm/kvm_host.h: atomic_t in kvm_s390_sie_block

drivers/net/wireless/ath/ath10k/core.h: dma_addr_t in ath10k_skb_cb
drivers/net/wireless/ath/ath11k/core.h: dma_addr_t in ath10k_skb_cb
drivers/crypto/ccp/ccp-dev.h: dma_addr_t in ccp_dma_info

The last 3 structures have a dma_adddr_t member defined as the first 
member variable. Most of the others also seems valid.

I used this SmPL script to find them:

@e1@
type T;
identifier i;
position p;
attribute name __packed;
@@
 T@p  {
  ...
(
  atomic_t i;
|
  raw_spinlock_t i;
|
  struct mutex i;
|
  spinlock_t i;
|
  dma_addr_t i;
|
  phys_addr_t i;
|
  size_t i;
|
 struct device i;
)
  ...
 }
 __packed;

@e2@
type T;
identifier i;
position p;
@@
 T@p  {
  ...
(
  atomic_t i;
|
  raw_spinlock_t i;
|
  struct mutex i;
|
  spinlock_t i;
|
  dma_addr_t i;
|
  phys_addr_t i;
|
  size_t i;
|
 struct device i;
)
  ...
 }
   __attribute__((
(
 pack
|
 __pack__
)
 ,... ));

@script:python@
ps <<e1.p;
@@
for p in ps:
  print p.file, p.line
@script:python@
ps <<e2.p;
@@
for p in ps:
  print p.file, p.line


Keisuke