netdev - Re: [PATCH 2/2] eventfd: simplify eventfd_signal

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]
Date: Thu, 13 Jul 2023 16:52:34 +0200
From: Christian Brauner <brauner@...nel.org>
To: Sean Christopherson <seanjc@...gle.com>
Cc: linux-fsdevel@...r.kernel.org, Vitaly Kuznetsov <vkuznets@...hat.com>,
	Paolo Bonzini <pbonzini@...hat.com>,
	Thomas Gleixner <tglx@...utronix.de>,
	Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
	Dave Hansen <dave.hansen@...ux.intel.com>, x86@...nel.org,
	David Woodhouse <dwmw2@...radead.org>, Paul Durrant <paul@....org>,
	Oded Gabbay <ogabbay@...nel.org>, Wu Hao <hao.wu@...el.com>,
	Tom Rix <trix@...hat.com>, Moritz Fischer <mdf@...nel.org>,
	Xu Yilun <yilun.xu@...el.com>,
	Zhenyu Wang <zhenyuw@...ux.intel.com>,
	Zhi Wang <zhi.a.wang@...el.com>,
	Jani Nikula <jani.nikula@...ux.intel.com>,
	Joonas Lahtinen <joonas.lahtinen@...ux.intel.com>,
	Rodrigo Vivi <rodrigo.vivi@...el.com>,
	Tvrtko Ursulin <tvrtko.ursulin@...ux.intel.com>,
	David Airlie <airlied@...il.com>, Daniel Vetter <daniel@...ll.ch>,
	Leon Romanovsky <leon@...nel.org>, Jason Gunthorpe <jgg@...pe.ca>,
	Frederic Barrat <fbarrat@...ux.ibm.com>,
	Andrew Donnellan <ajd@...ux.ibm.com>, Arnd Bergmann <arnd@...db.de>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Eric Farman <farman@...ux.ibm.com>,
	Matthew Rosato <mjrosato@...ux.ibm.com>,
	Halil Pasic <pasic@...ux.ibm.com>,
	Vineeth Vijayan <vneethv@...ux.ibm.com>,
	Peter Oberparleiter <oberpar@...ux.ibm.com>,
	Heiko Carstens <hca@...ux.ibm.com>,
	Vasily Gorbik <gor@...ux.ibm.com>,
	Alexander Gordeev <agordeev@...ux.ibm.com>,
	Christian Borntraeger <borntraeger@...ux.ibm.com>,
	Sven Schnelle <svens@...ux.ibm.com>,
	Tony Krowiak <akrowiak@...ux.ibm.com>,
	Jason Herne <jjherne@...ux.ibm.com>,
	Harald Freudenberger <freude@...ux.ibm.com>,
	"Michael S. Tsirkin" <mst@...hat.com>,
	Jason Wang <jasowang@...hat.com>,
	Xuan Zhuo <xuanzhuo@...ux.alibaba.com>,
	Diana Craciun <diana.craciun@....nxp.com>,
	Alex Williamson <alex.williamson@...hat.com>,
	Eric Auger <eric.auger@...hat.com>, Fei Li <fei1.li@...el.com>,
	Benjamin LaHaise <bcrl@...ck.org>,
	Johannes Weiner <hannes@...xchg.org>,
	Michal Hocko <mhocko@...nel.org>,
	Roman Gushchin <roman.gushchin@...ux.dev>,
	Shakeel Butt <shakeelb@...gle.com>,
	Muchun Song <muchun.song@...ux.dev>,
	Kirti Wankhede <kwankhede@...dia.com>, kvm@...r.kernel.org,
	linux-kernel@...r.kernel.org, dri-devel@...ts.freedesktop.org,
	linux-fpga@...r.kernel.org, intel-gvt-dev@...ts.freedesktop.org,
	intel-gfx@...ts.freedesktop.org, linux-rdma@...r.kernel.org,
	linuxppc-dev@...ts.ozlabs.org, linux-s390@...r.kernel.org,
	linux-usb@...r.kernel.org,
	virtualization@...ts.linux-foundation.org, netdev@...r.kernel.org,
	linux-aio@...ck.org, cgroups@...r.kernel.org, linux-mm@...ck.org,
	Jens Axboe <axboe@...nel.dk>,
	Pavel Begunkov <asml.silence@...il.com>, io-uring@...r.kernel.org
Subject: Re: [PATCH 2/2] eventfd: simplify eventfd_signal_mask()

On Thu, Jul 13, 2023 at 07:33:05AM -0700, Sean Christopherson wrote:
> On Thu, Jul 13, 2023, Christian Brauner wrote:
> > diff --git a/fs/eventfd.c b/fs/eventfd.c
> > index dc9e01053235..077be5da72bd 100644
> > --- a/fs/eventfd.c
> > +++ b/fs/eventfd.c
> > @@ -43,9 +43,10 @@ struct eventfd_ctx {
> >  	int id;
> >  };
> >  
> > -__u64 eventfd_signal_mask(struct eventfd_ctx *ctx, __u64 n, __poll_t mask)
> > +bool eventfd_signal_mask(struct eventfd_ctx *ctx, __poll_t mask)
> >  {
> >  	unsigned long flags;
> > +	__u64 n = 1;
> >  
> >  	/*
> >  	 * Deadlock or stack overflow issues can happen if we recurse here
> > @@ -68,7 +69,7 @@ __u64 eventfd_signal_mask(struct eventfd_ctx *ctx, __u64 n, __poll_t mask)
> >  	current->in_eventfd = 0;
> >  	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
> >  
> > -	return n;
> > +	return n == 1;
> >  }
> 
> ...
> 
> > @@ -58,13 +58,12 @@ static inline struct eventfd_ctx *eventfd_ctx_fdget(int fd)
> >  	return ERR_PTR(-ENOSYS);
> >  }
> >  
> > -static inline int eventfd_signal(struct eventfd_ctx *ctx)
> > +static inline bool eventfd_signal(struct eventfd_ctx *ctx)
> >  {
> >  	return -ENOSYS;
> >  }
> >  
> > -static inline int eventfd_signal_mask(struct eventfd_ctx *ctx, __u64 n,
> > -				      unsigned mask)
> > +static inline bool eventfd_signal_mask(struct eventfd_ctx *ctx, unsigned mask)
> >  {
> >  	return -ENOSYS;
> 
> This will morph to "true" for what should be an error case.  One option would be

Ewww, that means it did return -ENOSYS before any of this.

> to have eventfd_signal_mask() return 0/-errno instead of the count, but looking
> at all the callers, nothing ever actually consumes the result.
> 
> KVMGT morphs failure into -EFAULT
> 
> 	if (vgpu->msi_trigger && eventfd_signal(vgpu->msi_trigger, 1) != 1)
> 		return -EFAULT;
> 
> but the only caller of that user ignores the return value.
> 
> 	if (vgpu_vreg(vgpu, i915_mmio_reg_offset(GEN8_MASTER_IRQ))
> 			& ~GEN8_MASTER_IRQ_CONTROL)
> 		inject_virtual_interrupt(vgpu);
> 
> The sample driver in samples/vfio-mdev/mtty.c uses a similar pattern: prints an
> error but otherwise ignores the result.
> 
> So why not return nothing?  That will simplify eventfd_signal_mask() a wee bit
> more, and eliminate that bizarre return value confusion for the ugly stubs, e.g.

Yeah, it used to return an int in the non-eventfd and a __u64 in the
eventfd case.

> 
> void eventfd_signal_mask(struct eventfd_ctx *ctx, unsigned mask)
> {
> 	unsigned long flags;
> 
> 	/*
> 	 * Deadlock or stack overflow issues can happen if we recurse here
> 	 * through waitqueue wakeup handlers. If the caller users potentially
> 	 * nested waitqueues with custom wakeup handlers, then it should
> 	 * check eventfd_signal_allowed() before calling this function. If
> 	 * it returns false, the eventfd_signal() call should be deferred to a
> 	 * safe context.
> 	 */
> 	if (WARN_ON_ONCE(current->in_eventfd))
> 		return;
> 
> 	spin_lock_irqsave(&ctx->wqh.lock, flags);
> 	current->in_eventfd = 1;
> 	if (ctx->count < ULLONG_MAX)
> 		ctx->count++;
> 	if (waitqueue_active(&ctx->wqh))
> 		wake_up_locked_poll(&ctx->wqh, EPOLLIN | mask);
> 	current->in_eventfd = 0;
> 	spin_unlock_irqrestore(&ctx->wqh.lock, flags);
> }
> 
> You could even go further and unify the real and stub versions of eventfd_signal().

The reason I didn't make eventfd_signal_mask() return void was that it
was called from eventfd_signal() which did, I didn't realize the caller
didn't actually consume the return value.

If we can let both return void it gets simpler.

Thanks for that.