netdev - Re: [PATCH net-next v7 1/5] rust: core abstractions for network PHY drivers

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <8f476b7c-4647-457b-ab45-d6a979da4e78@lunn.ch>
Date: Fri, 10 Nov 2023 14:26:09 +0100
From: Andrew Lunn <andrew@...n.ch>
To: FUJITA Tomonori <fujita.tomonori@...il.com>
Cc: benno.lossin@...ton.me, boqun.feng@...il.com, netdev@...r.kernel.org,
	rust-for-linux@...r.kernel.org, tmgross@...ch.edu,
	miguel.ojeda.sandonis@...il.com, wedsonaf@...il.com
Subject: Re: [PATCH net-next v7 1/5] rust: core abstractions for network PHY
 drivers

On Wed, Nov 08, 2023 at 07:46:47PM +0900, FUJITA Tomonori wrote:
> On Mon, 30 Oct 2023 16:45:38 +0000
> Benno Lossin <benno.lossin@...ton.me> wrote:
> 
> >>> But I would wait until we see a response from the bindgen devs on the issue.
> >> 
> >> You meant that they might have a different option on this?
> > 
> > No, before you implement the workaround that Boqun posted you
> > should wait until the bindgen devs say how long/if they will
> > implement it.
> 
> It has been 10 days but no response from bindgen developpers. I guess
> that unlikely bindgen will support the feature until the next merge
> window.

I just took a look around the kernel includes. Here are a few examples
i found, there are many many more.

include/target/iscsi/iscsi_target_core.h:       unsigned int            conn_tx_reset_cpumask:1;
include/media/videobuf2-core.h: unsigned int            synced:1;
include/media/v4l2-ctrls.h:     unsigned int done:1;
include/drm/bridge/samsung-dsim.h:      unsigned int has_freqband:1;
include/net/sock_reuseport.h:   unsigned int            bind_inany:1;
include/net/sock.h:     unsigned char           skc_ipv6only:1;
include/sound/simple_card_utils.h:      unsigned int force_dpcm:1;
include/sound/soc.h:    unsigned int autodisable:1;
include/linux/pci.h:    unsigned int    imm_ready:1;    /* Supports Immediate Readiness */
include/linux/regmap.h: unsigned int use_ack:1;
include/linux/cpuidle.h:        unsigned int            registered:1;
include/linux/regulator/gpio-regulator.h:       unsigned enabled_at_boot:1;
include/linux/phy.h:    unsigned link:1;
include/linux/pm.h:     unsigned int            irq_safe:1;
include/linux/nfs_xdr.h:        unsigned char                   renew:1;
include/linux/iio/iio.h:        unsigned                output:1;
include/linux/drbd.h:           unsigned user_isp:1 ;
include/linux/sched.h:  unsigned                        sched_migrated:1;
include/linux/writeback.h:      unsigned no_cgroup_owner:1;
include/linux/tty_port.h:       unsigned char           console:1;
include/linux/mpi.h:                    unsigned int two_inv_p:1;
include/linux/mmc/host.h:       unsigned int            use_spi_crc:1;
include/linux/netdevice.h:      unsigned                wol_enabled:1;
include/linux/serial_8250.h:    unsigned int            tx_stopped:1;
include/linux/usb.h:    unsigned can_submit:1;
include/linux/firewire.h:       unsigned is_local:1;
include/linux/phylink.h:        unsigned int link:1;
include/linux/gpio_keys.h:      unsigned int rep:1;
include/linux/spi/spi.h:        unsigned        cs_change:1;
include/linux/i2c-mux.h:        unsigned int arbitrator:1;
include/linux/kobject.h:        unsigned int state_in_sysfs:1;
include/linux/kbd_kern.h:       unsigned char ledmode:1;
include/linux/bpf_verifier.h:   unsigned int fit_for_inline:1;
include/linux/rtc.h:    unsigned int uie_irq_active:1;
include/linux/usb/usbnet.h:     unsigned                can_dma_sg:1;
include/linux/usb/serial.h:     unsigned char                   attached:1;
include/linux/usb/gadget.h:     unsigned dir_out:1;
include/linux/comedi/comedidev.h:       unsigned int attached:1;
include/scsi/sg.h:    unsigned int twelve_byte:1;
include/scsi/scsi_host.h:       unsigned emulated:1;
include/scsi/scsi_device.h:     unsigned removable:1;
include/rdma/ib_verbs.h:        unsigned int            ip_gids:1;

And thats just bitfields used as binary values. There are more with
:2, :3, :4, :8, :16.

The point i'm trying to make is that they are used in many
subsystems. Not having bindgen support for them seems like its going
to be a problem.

Isn't this also an unsoundness problem? Is there existing Rust code
which people think is sound but is actually not?

> I prefer adding accessors in the C side rather than the workaround if
> it's fine by Andrew because we have no idea when bindgen will support
> the feature.

Maybe we need a better understanding of the kernel wide implications
of this. It could be this is actually a big issue, and rust-for-linux
can then apply either pressure, or human resources, to get it
implemented.

    Andrew