- Linux-stable-mirror - lists.linaro.org

Patch "fs/aio: Use RCU accessors for kioctx_table->table[]" has been added to the 3.18-stable tree

by gregkh＠linuxfoundation.org

This is a note to let you know that I've just added the patch titled fs/aio: Use RCU accessors for kioctx_table->table[] to the 3.18-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum… The filename of the patch is: fs-aio-use-rcu-accessors-for-kioctx_table-table.patch and it can be found in the queue-3.18 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable(a)vger.kernel.org> know about it. >From d0264c01e7587001a8c4608a5d1818dba9a4c11a Mon Sep 17 00:00:00 2001 From: Tejun Heo <tj(a)kernel.org> Date: Wed, 14 Mar 2018 12:10:17 -0700 Subject: fs/aio: Use RCU accessors for kioctx_table->table[] From: Tejun Heo <tj(a)kernel.org> commit d0264c01e7587001a8c4608a5d1818dba9a4c11a upstream. While converting ioctx index from a list to a table, db446a08c23d ("aio: convert the ioctx list to table lookup v3") missed tagging kioctx_table->table[] as an array of RCU pointers and using the appropriate RCU accessors. This introduces a small window in the lookup path where init and access may race. Mark kioctx_table->table[] with __rcu and use the approriate RCU accessors when using the field. Signed-off-by: Tejun Heo <tj(a)kernel.org> Reported-by: Jann Horn <jannh(a)google.com> Fixes: db446a08c23d ("aio: convert the ioctx list to table lookup v3") Cc: Benjamin LaHaise <bcrl(a)kvack.org> Cc: Linus Torvalds <torvalds(a)linux-foundation.org> Cc: stable(a)vger.kernel.org # v3.12+ Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> --- fs/aio.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) --- a/fs/aio.c +++ b/fs/aio.c @@ -68,9 +68,9 @@ struct aio_ring { #define AIO_RING_PAGES 8 struct kioctx_table { - struct rcu_head rcu; - unsigned nr; - struct kioctx *table[]; + struct rcu_head rcu; + unsigned nr; + struct kioctx __rcu *table[]; }; struct kioctx_cpu { @@ -583,9 +583,9 @@ static int ioctx_add_table(struct kioctx while (1) { if (table) for (i = 0; i < table->nr; i++) - if (!table->table[i]) { + if (!rcu_access_pointer(table->table[i])) { ctx->id = i; - table->table[i] = ctx; + rcu_assign_pointer(table->table[i], ctx); spin_unlock(&mm->ioctx_lock); /* While kioctx setup is in progress, @@ -760,8 +760,8 @@ static int kill_ioctx(struct mm_struct * spin_lock(&mm->ioctx_lock); table = rcu_dereference_raw(mm->ioctx_table); - WARN_ON(ctx != table->table[ctx->id]); - table->table[ctx->id] = NULL; + WARN_ON(ctx != rcu_access_pointer(table->table[ctx->id])); + RCU_INIT_POINTER(table->table[ctx->id], NULL); spin_unlock(&mm->ioctx_lock); /* free_ioctx_reqs() will do the necessary RCU synchronization */ @@ -817,7 +817,8 @@ void exit_aio(struct mm_struct *mm) return; for (i = 0; i < table->nr; ++i) { - struct kioctx *ctx = table->table[i]; + struct kioctx *ctx = + rcu_dereference_protected(table->table[i], true); struct completion requests_done = COMPLETION_INITIALIZER_ONSTACK(requests_done); @@ -1003,7 +1004,7 @@ static struct kioctx *lookup_ioctx(unsig if (!table || id >= table->nr) goto out; - ctx = table->table[id]; + ctx = rcu_dereference(table->table[id]); if (ctx && ctx->user_id == ctx_id) { percpu_ref_get(&ctx->users); ret = ctx; Patches currently in stable-queue which might be from tj(a)kernel.org are queue-3.18/fs-aio-add-explicit-rcu-grace-period-when-freeing-kioctx.patch queue-3.18/fs-aio-use-rcu-accessors-for-kioctx_table-table.patch

7 years, 3 months

1
0
0 0

Patch "fs/aio: Add explicit RCU grace period when freeing kioctx" has been added to the 3.18-stable tree

by gregkh＠linuxfoundation.org

This is a note to let you know that I've just added the patch titled fs/aio: Add explicit RCU grace period when freeing kioctx to the 3.18-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum… The filename of the patch is: fs-aio-add-explicit-rcu-grace-period-when-freeing-kioctx.patch and it can be found in the queue-3.18 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable(a)vger.kernel.org> know about it. >From a6d7cff472eea87d96899a20fa718d2bab7109f3 Mon Sep 17 00:00:00 2001 From: Tejun Heo <tj(a)kernel.org> Date: Wed, 14 Mar 2018 12:10:17 -0700 Subject: fs/aio: Add explicit RCU grace period when freeing kioctx From: Tejun Heo <tj(a)kernel.org> commit a6d7cff472eea87d96899a20fa718d2bab7109f3 upstream. While fixing refcounting, e34ecee2ae79 ("aio: Fix a trinity splat") incorrectly removed explicit RCU grace period before freeing kioctx. The intention seems to be depending on the internal RCU grace periods of percpu_ref; however, percpu_ref uses a different flavor of RCU, sched-RCU. This can lead to kioctx being freed while RCU read protected dereferences are still in progress. Fix it by updating free_ioctx() to go through call_rcu() explicitly. v2: Comment added to explain double bouncing. Signed-off-by: Tejun Heo <tj(a)kernel.org> Reported-by: Jann Horn <jannh(a)google.com> Fixes: e34ecee2ae79 ("aio: Fix a trinity splat") Cc: Kent Overstreet <kent.overstreet(a)gmail.com> Cc: Linus Torvalds <torvalds(a)linux-foundation.org> Cc: stable(a)vger.kernel.org # v3.13+ Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> --- fs/aio.c | 23 +++++++++++++++++++---- 1 file changed, 19 insertions(+), 4 deletions(-) --- a/fs/aio.c +++ b/fs/aio.c @@ -110,7 +110,8 @@ struct kioctx { struct page **ring_pages; long nr_pages; - struct work_struct free_work; + struct rcu_head free_rcu; + struct work_struct free_work; /* see free_ioctx() */ /* * signals when all in-flight requests are done @@ -505,6 +506,12 @@ static int kiocb_cancel(struct kiocb *ki return cancel(kiocb); } +/* + * free_ioctx() should be RCU delayed to synchronize against the RCU + * protected lookup_ioctx() and also needs process context to call + * aio_free_ring(), so the double bouncing through kioctx->free_rcu and + * ->free_work. + */ static void free_ioctx(struct work_struct *work) { struct kioctx *ctx = container_of(work, struct kioctx, free_work); @@ -518,6 +525,14 @@ static void free_ioctx(struct work_struc kmem_cache_free(kioctx_cachep, ctx); } +static void free_ioctx_rcufn(struct rcu_head *head) +{ + struct kioctx *ctx = container_of(head, struct kioctx, free_rcu); + + INIT_WORK(&ctx->free_work, free_ioctx); + schedule_work(&ctx->free_work); +} + static void free_ioctx_reqs(struct percpu_ref *ref) { struct kioctx *ctx = container_of(ref, struct kioctx, reqs); @@ -526,8 +541,8 @@ static void free_ioctx_reqs(struct percp if (ctx->requests_done) complete(ctx->requests_done); - INIT_WORK(&ctx->free_work, free_ioctx); - schedule_work(&ctx->free_work); + /* Synchronize against RCU protected table->table[] dereferences */ + call_rcu(&ctx->free_rcu, free_ioctx_rcufn); } /* @@ -749,7 +764,7 @@ static int kill_ioctx(struct mm_struct * table->table[ctx->id] = NULL; spin_unlock(&mm->ioctx_lock); - /* percpu_ref_kill() will do the necessary call_rcu() */ + /* free_ioctx_reqs() will do the necessary RCU synchronization */ wake_up_all(&ctx->wait); /* Patches currently in stable-queue which might be from tj(a)kernel.org are queue-3.18/fs-aio-add-explicit-rcu-grace-period-when-freeing-kioctx.patch queue-3.18/fs-aio-use-rcu-accessors-for-kioctx_table-table.patch

7 years, 3 months

1
0
0 0

Patch "ALSA: seq: Fix possible UAF in snd_seq_check_queue()" has been added to the 3.18-stable tree

by gregkh＠linuxfoundation.org

This is a note to let you know that I've just added the patch titled ALSA: seq: Fix possible UAF in snd_seq_check_queue() to the 3.18-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum… The filename of the patch is: alsa-seq-fix-possible-uaf-in-snd_seq_check_queue.patch and it can be found in the queue-3.18 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable(a)vger.kernel.org> know about it. >From d0f833065221cbfcbadf19fd4102bcfa9330006a Mon Sep 17 00:00:00 2001 From: Takashi Iwai <tiwai(a)suse.de> Date: Fri, 9 Mar 2018 21:58:28 +0100 Subject: ALSA: seq: Fix possible UAF in snd_seq_check_queue() From: Takashi Iwai <tiwai(a)suse.de> commit d0f833065221cbfcbadf19fd4102bcfa9330006a upstream. Although we've covered the races between concurrent write() and ioctl() in the previous patch series, there is still a possible UAF in the following scenario: A: user client closed B: timer irq -> snd_seq_release() -> snd_seq_timer_interrupt() -> snd_seq_free_client() -> snd_seq_check_queue() -> cell = snd_seq_prioq_cell_peek() -> snd_seq_prioq_leave() .... removing all cells -> snd_seq_pool_done() .... vfree() -> snd_seq_compare_tick_time(cell) ... Oops So the problem is that a cell is peeked and accessed without any protection until it's retrieved from the queue again via snd_seq_prioq_cell_out(). This patch tries to address it, also cleans up the code by a slight refactoring. snd_seq_prioq_cell_out() now receives an extra pointer argument. When it's non-NULL, the function checks the event timestamp with the given pointer. The caller needs to pass the right reference either to snd_seq_tick or snd_seq_realtime depending on the event timestamp type. A good news is that the above change allows us to remove the snd_seq_prioq_cell_peek(), too, thus the patch actually reduces the code size. Reviewed-by: Nicolai Stange <nstange(a)suse.de> Cc: <stable(a)vger.kernel.org> Signed-off-by: Takashi Iwai <tiwai(a)suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> --- sound/core/seq/seq_prioq.c | 28 ++++++++++++++-------------- sound/core/seq/seq_prioq.h | 6 ++---- sound/core/seq/seq_queue.c | 28 +++++++++------------------- 3 files changed, 25 insertions(+), 37 deletions(-) --- a/sound/core/seq/seq_prioq.c +++ b/sound/core/seq/seq_prioq.c @@ -87,7 +87,7 @@ void snd_seq_prioq_delete(struct snd_seq if (f->cells > 0) { /* drain prioQ */ while (f->cells > 0) - snd_seq_cell_free(snd_seq_prioq_cell_out(f)); + snd_seq_cell_free(snd_seq_prioq_cell_out(f, NULL)); } kfree(f); @@ -214,8 +214,18 @@ int snd_seq_prioq_cell_in(struct snd_seq return 0; } +/* return 1 if the current time >= event timestamp */ +static int event_is_ready(struct snd_seq_event *ev, void *current_time) +{ + if ((ev->flags & SNDRV_SEQ_TIME_STAMP_MASK) == SNDRV_SEQ_TIME_STAMP_TICK) + return snd_seq_compare_tick_time(current_time, &ev->time.tick); + else + return snd_seq_compare_real_time(current_time, &ev->time.time); +} + /* dequeue cell from prioq */ -struct snd_seq_event_cell *snd_seq_prioq_cell_out(struct snd_seq_prioq *f) +struct snd_seq_event_cell *snd_seq_prioq_cell_out(struct snd_seq_prioq *f, + void *current_time) { struct snd_seq_event_cell *cell; unsigned long flags; @@ -227,6 +237,8 @@ struct snd_seq_event_cell *snd_seq_prioq spin_lock_irqsave(&f->lock, flags); cell = f->head; + if (cell && current_time && !event_is_ready(&cell->event, current_time)) + cell = NULL; if (cell) { f->head = cell->next; @@ -252,18 +264,6 @@ int snd_seq_prioq_avail(struct snd_seq_p return f->cells; } - -/* peek at cell at the head of the prioq */ -struct snd_seq_event_cell *snd_seq_prioq_cell_peek(struct snd_seq_prioq * f) -{ - if (f == NULL) { - pr_debug("ALSA: seq: snd_seq_prioq_cell_in() called with NULL prioq\n"); - return NULL; - } - return f->head; -} - - static inline int prioq_match(struct snd_seq_event_cell *cell, int client, int timestamp) { --- a/sound/core/seq/seq_prioq.h +++ b/sound/core/seq/seq_prioq.h @@ -44,14 +44,12 @@ void snd_seq_prioq_delete(struct snd_seq int snd_seq_prioq_cell_in(struct snd_seq_prioq *f, struct snd_seq_event_cell *cell); /* dequeue cell from prioq */ -struct snd_seq_event_cell *snd_seq_prioq_cell_out(struct snd_seq_prioq *f); +struct snd_seq_event_cell *snd_seq_prioq_cell_out(struct snd_seq_prioq *f, + void *current_time); /* return number of events available in prioq */ int snd_seq_prioq_avail(struct snd_seq_prioq *f); -/* peek at cell at the head of the prioq */ -struct snd_seq_event_cell *snd_seq_prioq_cell_peek(struct snd_seq_prioq *f); - /* client left queue */ void snd_seq_prioq_leave(struct snd_seq_prioq *f, int client, int timestamp); --- a/sound/core/seq/seq_queue.c +++ b/sound/core/seq/seq_queue.c @@ -271,30 +271,20 @@ void snd_seq_check_queue(struct snd_seq_ __again: /* Process tick queue... */ - while ((cell = snd_seq_prioq_cell_peek(q->tickq)) != NULL) { - if (snd_seq_compare_tick_time(&q->timer->tick.cur_tick, - &cell->event.time.tick)) { - cell = snd_seq_prioq_cell_out(q->tickq); - if (cell) - snd_seq_dispatch_event(cell, atomic, hop); - } else { - /* event remains in the queue */ + for (;;) { + cell = snd_seq_prioq_cell_out(q->tickq, + &q->timer->tick.cur_tick); + if (!cell) break; - } + snd_seq_dispatch_event(cell, atomic, hop); } - /* Process time queue... */ - while ((cell = snd_seq_prioq_cell_peek(q->timeq)) != NULL) { - if (snd_seq_compare_real_time(&q->timer->cur_time, - &cell->event.time.time)) { - cell = snd_seq_prioq_cell_out(q->timeq); - if (cell) - snd_seq_dispatch_event(cell, atomic, hop); - } else { - /* event remains in the queue */ + for (;;) { + cell = snd_seq_prioq_cell_out(q->timeq, &q->timer->cur_time); + if (!cell) break; - } + snd_seq_dispatch_event(cell, atomic, hop); } /* free lock */ Patches currently in stable-queue which might be from tiwai(a)suse.de are queue-3.18/alsa-pcm-fix-uaf-in-snd_pcm_oss_get_formats.patch queue-3.18/alsa-seq-fix-possible-uaf-in-snd_seq_check_queue.patch queue-3.18/alsa-seq-clear-client-entry-before-deleting-else-at-closing.patch

7 years, 3 months

1
0
0 0

Patch "ALSA: seq: Clear client entry before deleting else at closing" has been added to the 3.18-stable tree

by gregkh＠linuxfoundation.org

This is a note to let you know that I've just added the patch titled ALSA: seq: Clear client entry before deleting else at closing to the 3.18-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum… The filename of the patch is: alsa-seq-clear-client-entry-before-deleting-else-at-closing.patch and it can be found in the queue-3.18 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable(a)vger.kernel.org> know about it. >From a2ff19f7b70118ced291a28d5313469914de451b Mon Sep 17 00:00:00 2001 From: Takashi Iwai <tiwai(a)suse.de> Date: Fri, 9 Mar 2018 22:23:31 +0100 Subject: ALSA: seq: Clear client entry before deleting else at closing From: Takashi Iwai <tiwai(a)suse.de> commit a2ff19f7b70118ced291a28d5313469914de451b upstream. When releasing a client, we need to clear the clienttab[] entry at first, then call snd_seq_queue_client_leave(). Otherwise, the in-flight cell in the queue might be picked up by the timer interrupt via snd_seq_check_queue() before calling snd_seq_queue_client_leave(), and it's delivered to another queue while the client is clearing queues. This may eventually result in an uncleared cell remaining in a queue, and the later snd_seq_pool_delete() may need to wait for a long time until the event gets really processed. By moving the clienttab[] clearance at the beginning of release, any event delivery of a cell belonging to this client will fail at a later point, since snd_seq_client_ptr() returns NULL. Thus the cell that was picked up by the timer interrupt will be returned immediately without further delivery, and the long stall of snd_seq_delete_pool() can be avoided, too. Cc: <stable(a)vger.kernel.org> Signed-off-by: Takashi Iwai <tiwai(a)suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> --- sound/core/seq/seq_clientmgr.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) --- a/sound/core/seq/seq_clientmgr.c +++ b/sound/core/seq/seq_clientmgr.c @@ -270,12 +270,12 @@ static int seq_free_client1(struct snd_s if (!client) return 0; - snd_seq_delete_all_ports(client); - snd_seq_queue_client_leave(client->number); spin_lock_irqsave(&clients_lock, flags); clienttablock[client->number] = 1; clienttab[client->number] = NULL; spin_unlock_irqrestore(&clients_lock, flags); + snd_seq_delete_all_ports(client); + snd_seq_queue_client_leave(client->number); snd_use_lock_sync(&client->use_lock); snd_seq_queue_client_termination(client->number); if (client->pool) Patches currently in stable-queue which might be from tiwai(a)suse.de are queue-3.18/alsa-pcm-fix-uaf-in-snd_pcm_oss_get_formats.patch queue-3.18/alsa-seq-fix-possible-uaf-in-snd_seq_check_queue.patch queue-3.18/alsa-seq-clear-client-entry-before-deleting-else-at-closing.patch

7 years, 3 months

1
0
0 0

Patch "ALSA: pcm: Fix UAF in snd_pcm_oss_get_formats()" has been added to the 3.18-stable tree

by gregkh＠linuxfoundation.org

This is a note to let you know that I've just added the patch titled ALSA: pcm: Fix UAF in snd_pcm_oss_get_formats() to the 3.18-stable tree which can be found at: http://www.kernel.org/git/?p=linux/kernel/git/stable/stable-queue.git;a=sum… The filename of the patch is: alsa-pcm-fix-uaf-in-snd_pcm_oss_get_formats.patch and it can be found in the queue-3.18 subdirectory. If you, or anyone else, feels it should not be added to the stable tree, please let <stable(a)vger.kernel.org> know about it. >From 01c0b4265cc16bc1f43f475c5944c55c10d5768f Mon Sep 17 00:00:00 2001 From: Takashi Iwai <tiwai(a)suse.de> Date: Sat, 10 Mar 2018 23:04:23 +0100 Subject: ALSA: pcm: Fix UAF in snd_pcm_oss_get_formats() From: Takashi Iwai <tiwai(a)suse.de> commit 01c0b4265cc16bc1f43f475c5944c55c10d5768f upstream. snd_pcm_oss_get_formats() has an obvious use-after-free around snd_mask_test() calls, as spotted by syzbot. The passed format_mask argument is a pointer to the hw_params object that is freed before the loop. What a surprise that it has been present since the original code of decades ago... Reported-by: syzbot+4090700a4f13fccaf648(a)syzkaller.appspotmail.com Cc: <stable(a)vger.kernel.org> Signed-off-by: Takashi Iwai <tiwai(a)suse.de> Signed-off-by: Greg Kroah-Hartman <gregkh(a)linuxfoundation.org> --- sound/core/oss/pcm_oss.c | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) --- a/sound/core/oss/pcm_oss.c +++ b/sound/core/oss/pcm_oss.c @@ -1815,10 +1815,9 @@ static int snd_pcm_oss_get_formats(struc return -ENOMEM; _snd_pcm_hw_params_any(params); err = snd_pcm_hw_refine(substream, params); - format_mask = *hw_param_mask(params, SNDRV_PCM_HW_PARAM_FORMAT); - kfree(params); if (err < 0) - return err; + goto error; + format_mask = hw_param_mask_c(params, SNDRV_PCM_HW_PARAM_FORMAT); for (fmt = 0; fmt < 32; ++fmt) { if (snd_mask_test(&format_mask, fmt)) { int f = snd_pcm_oss_format_to(fmt); @@ -1826,7 +1825,10 @@ static int snd_pcm_oss_get_formats(struc formats |= f; } } - return formats; + + error: + kfree(params); + return err < 0 ? err : formats; } static int snd_pcm_oss_set_format(struct snd_pcm_oss_file *pcm_oss_file, int format) Patches currently in stable-queue which might be from tiwai(a)suse.de are queue-3.18/alsa-pcm-fix-uaf-in-snd_pcm_oss_get_formats.patch queue-3.18/alsa-seq-fix-possible-uaf-in-snd_seq_check_queue.patch queue-3.18/alsa-seq-clear-client-entry-before-deleting-else-at-closing.patch

7 years, 3 months

1
0
0 0

FAILED: patch "[PATCH] KVM: arm/arm64: vgic: Don't populate multiple LRs with the" failed to apply to 4.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 16ca6a607d84bef0129698d8d808f501afd08d43 Mon Sep 17 00:00:00 2001 From: Marc Zyngier <marc.zyngier(a)arm.com> Date: Tue, 6 Mar 2018 21:48:01 +0000 Subject: [PATCH] KVM: arm/arm64: vgic: Don't populate multiple LRs with the same vintid The vgic code is trying to be clever when injecting GICv2 SGIs, and will happily populate LRs with the same interrupt number if they come from multiple vcpus (after all, they are distinct interrupt sources). Unfortunately, this is against the letter of the architecture, and the GICv2 architecture spec says "Each valid interrupt stored in the List registers must have a unique VirtualID for that virtual CPU interface.". GICv3 has similar (although slightly ambiguous) restrictions. This results in guests locking up when using GICv2-on-GICv3, for example. The obvious fix is to stop trying so hard, and inject a single vcpu per SGI per guest entry. After all, pending SGIs with multiple source vcpus are pretty rare, and are mostly seen in scenario where the physical CPUs are severely overcomitted. But as we now only inject a single instance of a multi-source SGI per vcpu entry, we may delay those interrupts for longer than strictly necessary, and run the risk of injecting lower priority interrupts in the meantime. In order to address this, we adopt a three stage strategy: - If we encounter a multi-source SGI in the AP list while computing its depth, we force the list to be sorted - When populating the LRs, we prevent the injection of any interrupt of lower priority than that of the first multi-source SGI we've injected. - Finally, the injection of a multi-source SGI triggers the request of a maintenance interrupt when there will be no pending interrupt in the LRs (HCR_NPIE). At the point where the last pending interrupt in the LRs switches from Pending to Active, the maintenance interrupt will be delivered, allowing us to add the remaining SGIs using the same process. Cc: stable(a)vger.kernel.org Fixes: 0919e84c0fc1 ("KVM: arm/arm64: vgic-new: Add IRQ sync/flush framework") Acked-by: Christoffer Dall <cdall(a)kernel.org> Signed-off-by: Marc Zyngier <marc.zyngier(a)arm.com> diff --git a/include/linux/irqchip/arm-gic-v3.h b/include/linux/irqchip/arm-gic-v3.h index c00c4c33e432..b26eccc78fb1 100644 --- a/include/linux/irqchip/arm-gic-v3.h +++ b/include/linux/irqchip/arm-gic-v3.h @@ -503,6 +503,7 @@ #define ICH_HCR_EN (1 << 0) #define ICH_HCR_UIE (1 << 1) +#define ICH_HCR_NPIE (1 << 3) #define ICH_HCR_TC (1 << 10) #define ICH_HCR_TALL0 (1 << 11) #define ICH_HCR_TALL1 (1 << 12) diff --git a/include/linux/irqchip/arm-gic.h b/include/linux/irqchip/arm-gic.h index d3453ee072fc..68d8b1f73682 100644 --- a/include/linux/irqchip/arm-gic.h +++ b/include/linux/irqchip/arm-gic.h @@ -84,6 +84,7 @@ #define GICH_HCR_EN (1 << 0) #define GICH_HCR_UIE (1 << 1) +#define GICH_HCR_NPIE (1 << 3) #define GICH_LR_VIRTUALID (0x3ff << 0) #define GICH_LR_PHYSID_CPUID_SHIFT (10) diff --git a/virt/kvm/arm/vgic/vgic-v2.c b/virt/kvm/arm/vgic/vgic-v2.c index e9d840a75e7b..29556f71b691 100644 --- a/virt/kvm/arm/vgic/vgic-v2.c +++ b/virt/kvm/arm/vgic/vgic-v2.c @@ -37,6 +37,13 @@ void vgic_v2_init_lrs(void) vgic_v2_write_lr(i, 0); } +void vgic_v2_set_npie(struct kvm_vcpu *vcpu) +{ + struct vgic_v2_cpu_if *cpuif = &vcpu->arch.vgic_cpu.vgic_v2; + + cpuif->vgic_hcr |= GICH_HCR_NPIE; +} + void vgic_v2_set_underflow(struct kvm_vcpu *vcpu) { struct vgic_v2_cpu_if *cpuif = &vcpu->arch.vgic_cpu.vgic_v2; @@ -64,7 +71,7 @@ void vgic_v2_fold_lr_state(struct kvm_vcpu *vcpu) int lr; unsigned long flags; - cpuif->vgic_hcr &= ~GICH_HCR_UIE; + cpuif->vgic_hcr &= ~(GICH_HCR_UIE | GICH_HCR_NPIE); for (lr = 0; lr < vgic_cpu->used_lrs; lr++) { u32 val = cpuif->vgic_lr[lr]; diff --git a/virt/kvm/arm/vgic/vgic-v3.c b/virt/kvm/arm/vgic/vgic-v3.c index 6b329414e57a..0ff2006f3781 100644 --- a/virt/kvm/arm/vgic/vgic-v3.c +++ b/virt/kvm/arm/vgic/vgic-v3.c @@ -26,6 +26,13 @@ static bool group1_trap; static bool common_trap; static bool gicv4_enable; +void vgic_v3_set_npie(struct kvm_vcpu *vcpu) +{ + struct vgic_v3_cpu_if *cpuif = &vcpu->arch.vgic_cpu.vgic_v3; + + cpuif->vgic_hcr |= ICH_HCR_NPIE; +} + void vgic_v3_set_underflow(struct kvm_vcpu *vcpu) { struct vgic_v3_cpu_if *cpuif = &vcpu->arch.vgic_cpu.vgic_v3; @@ -47,7 +54,7 @@ void vgic_v3_fold_lr_state(struct kvm_vcpu *vcpu) int lr; unsigned long flags; - cpuif->vgic_hcr &= ~ICH_HCR_UIE; + cpuif->vgic_hcr &= ~(ICH_HCR_UIE | ICH_HCR_NPIE); for (lr = 0; lr < vgic_cpu->used_lrs; lr++) { u64 val = cpuif->vgic_lr[lr]; diff --git a/virt/kvm/arm/vgic/vgic.c b/virt/kvm/arm/vgic/vgic.c index 0001858a2c23..8201899126f6 100644 --- a/virt/kvm/arm/vgic/vgic.c +++ b/virt/kvm/arm/vgic/vgic.c @@ -710,22 +710,37 @@ static inline void vgic_set_underflow(struct kvm_vcpu *vcpu) vgic_v3_set_underflow(vcpu); } +static inline void vgic_set_npie(struct kvm_vcpu *vcpu) +{ + if (kvm_vgic_global_state.type == VGIC_V2) + vgic_v2_set_npie(vcpu); + else + vgic_v3_set_npie(vcpu); +} + /* Requires the ap_list_lock to be held. */ -static int compute_ap_list_depth(struct kvm_vcpu *vcpu) +static int compute_ap_list_depth(struct kvm_vcpu *vcpu, + bool *multi_sgi) { struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu; struct vgic_irq *irq; int count = 0; + *multi_sgi = false; + DEBUG_SPINLOCK_BUG_ON(!spin_is_locked(&vgic_cpu->ap_list_lock)); list_for_each_entry(irq, &vgic_cpu->ap_list_head, ap_list) { spin_lock(&irq->irq_lock); /* GICv2 SGIs can count for more than one... */ - if (vgic_irq_is_sgi(irq->intid) && irq->source) - count += hweight8(irq->source); - else + if (vgic_irq_is_sgi(irq->intid) && irq->source) { + int w = hweight8(irq->source); + + count += w; + *multi_sgi |= (w > 1); + } else { count++; + } spin_unlock(&irq->irq_lock); } return count; @@ -736,28 +751,43 @@ static void vgic_flush_lr_state(struct kvm_vcpu *vcpu) { struct vgic_cpu *vgic_cpu = &vcpu->arch.vgic_cpu; struct vgic_irq *irq; - int count = 0; + int count; + bool npie = false; + bool multi_sgi; + u8 prio = 0xff; DEBUG_SPINLOCK_BUG_ON(!spin_is_locked(&vgic_cpu->ap_list_lock)); - if (compute_ap_list_depth(vcpu) > kvm_vgic_global_state.nr_lr) + count = compute_ap_list_depth(vcpu, &multi_sgi); + if (count > kvm_vgic_global_state.nr_lr || multi_sgi) vgic_sort_ap_list(vcpu); + count = 0; + list_for_each_entry(irq, &vgic_cpu->ap_list_head, ap_list) { spin_lock(&irq->irq_lock); - if (unlikely(vgic_target_oracle(irq) != vcpu)) - goto next; - /* - * If we get an SGI with multiple sources, try to get - * them in all at once. + * If we have multi-SGIs in the pipeline, we need to + * guarantee that they are all seen before any IRQ of + * lower priority. In that case, we need to filter out + * these interrupts by exiting early. This is easy as + * the AP list has been sorted already. */ - do { + if (multi_sgi && irq->priority > prio) { + spin_unlock(&irq->irq_lock); + break; + } + + if (likely(vgic_target_oracle(irq) == vcpu)) { vgic_populate_lr(vcpu, irq, count++); - } while (irq->source && count < kvm_vgic_global_state.nr_lr); -next: + if (irq->source) { + npie = true; + prio = irq->priority; + } + } + spin_unlock(&irq->irq_lock); if (count == kvm_vgic_global_state.nr_lr) { @@ -768,6 +798,9 @@ static void vgic_flush_lr_state(struct kvm_vcpu *vcpu) } } + if (npie) + vgic_set_npie(vcpu); + vcpu->arch.vgic_cpu.used_lrs = count; /* Nuke remaining LRs */ diff --git a/virt/kvm/arm/vgic/vgic.h b/virt/kvm/arm/vgic/vgic.h index 5b11859a1a1e..f5b8519e5546 100644 --- a/virt/kvm/arm/vgic/vgic.h +++ b/virt/kvm/arm/vgic/vgic.h @@ -160,6 +160,7 @@ void vgic_v2_fold_lr_state(struct kvm_vcpu *vcpu); void vgic_v2_populate_lr(struct kvm_vcpu *vcpu, struct vgic_irq *irq, int lr); void vgic_v2_clear_lr(struct kvm_vcpu *vcpu, int lr); void vgic_v2_set_underflow(struct kvm_vcpu *vcpu); +void vgic_v2_set_npie(struct kvm_vcpu *vcpu); int vgic_v2_has_attr_regs(struct kvm_device *dev, struct kvm_device_attr *attr); int vgic_v2_dist_uaccess(struct kvm_vcpu *vcpu, bool is_write, int offset, u32 *val); @@ -189,6 +190,7 @@ void vgic_v3_fold_lr_state(struct kvm_vcpu *vcpu); void vgic_v3_populate_lr(struct kvm_vcpu *vcpu, struct vgic_irq *irq, int lr); void vgic_v3_clear_lr(struct kvm_vcpu *vcpu, int lr); void vgic_v3_set_underflow(struct kvm_vcpu *vcpu); +void vgic_v3_set_npie(struct kvm_vcpu *vcpu); void vgic_v3_set_vmcr(struct kvm_vcpu *vcpu, struct vgic_vmcr *vmcr); void vgic_v3_get_vmcr(struct kvm_vcpu *vcpu, struct vgic_vmcr *vmcr); void vgic_v3_enable(struct kvm_vcpu *vcpu);

7 years, 3 months

1
0
0 0

FAILED: patch "[PATCH] kvm: arm/arm64: vgic-v3: Tighten synchronization for guests" failed to apply to 4.9-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.9-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 27e91ad1e746e341ca2312f29bccb9736be7b476 Mon Sep 17 00:00:00 2001 From: Marc Zyngier <marc.zyngier(a)arm.com> Date: Tue, 6 Mar 2018 21:44:37 +0000 Subject: [PATCH] kvm: arm/arm64: vgic-v3: Tighten synchronization for guests using v2 on v3 On guest exit, and when using GICv2 on GICv3, we use a dsb(st) to force synchronization between the memory-mapped guest view and the system-register view that the hypervisor uses. This is incorrect, as the spec calls out the need for "a DSB whose required access type is both loads and stores with any Shareability attribute", while we're only synchronizing stores. We also lack an isb after the dsb to ensure that the latter has actually been executed before we start reading stuff from the sysregs. The fix is pretty easy: turn dsb(st) into dsb(sy), and slap an isb() just after. Cc: stable(a)vger.kernel.org Fixes: f68d2b1b73cc ("arm64: KVM: Implement vgic-v3 save/restore") Acked-by: Christoffer Dall <cdall(a)kernel.org> Reviewed-by: Andre Przywara <andre.przywara(a)arm.com> Signed-off-by: Marc Zyngier <marc.zyngier(a)arm.com> diff --git a/virt/kvm/arm/hyp/vgic-v3-sr.c b/virt/kvm/arm/hyp/vgic-v3-sr.c index f5c3d6d7019e..b89ce5432214 100644 --- a/virt/kvm/arm/hyp/vgic-v3-sr.c +++ b/virt/kvm/arm/hyp/vgic-v3-sr.c @@ -215,7 +215,8 @@ void __hyp_text __vgic_v3_save_state(struct kvm_vcpu *vcpu) * are now visible to the system register interface. */ if (!cpu_if->vgic_sre) { - dsb(st); + dsb(sy); + isb(); cpu_if->vgic_vmcr = read_gicreg(ICH_VMCR_EL2); }

7 years, 3 months

1
0
0 0

FAILED: patch "[PATCH] KVM: arm/arm64: Reset mapped IRQs on VM reset" failed to apply to 4.15-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.15-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From 413aa807ae39fed7e387c175d2d0ae9fcf6c0c9d Mon Sep 17 00:00:00 2001 From: Christoffer Dall <cdall(a)kernel.org> Date: Mon, 5 Mar 2018 11:36:38 +0100 Subject: [PATCH] KVM: arm/arm64: Reset mapped IRQs on VM reset We currently don't allow resetting mapped IRQs from userspace, because their state is controlled by the hardware. But we do need to reset the state when the VM is reset, so we provide a function for the 'owner' of the mapped interrupt to reset the interrupt state. Currently only the timer uses mapped interrupts, so we call this function from the timer reset logic. Cc: stable(a)vger.kernel.org Fixes: 4c60e360d6df ("KVM: arm/arm64: Provide a get_input_level for the arch timer") Signed-off-by: Christoffer Dall <cdall(a)kernel.org> Signed-off-by: Marc Zyngier <marc.zyngier(a)arm.com> diff --git a/include/kvm/arm_vgic.h b/include/kvm/arm_vgic.h index cdbd142ca7f2..02924ae2527e 100644 --- a/include/kvm/arm_vgic.h +++ b/include/kvm/arm_vgic.h @@ -360,6 +360,7 @@ void kvm_vgic_put(struct kvm_vcpu *vcpu); bool kvm_vcpu_has_pending_irqs(struct kvm_vcpu *vcpu); void kvm_vgic_sync_hwstate(struct kvm_vcpu *vcpu); void kvm_vgic_flush_hwstate(struct kvm_vcpu *vcpu); +void kvm_vgic_reset_mapped_irq(struct kvm_vcpu *vcpu, u32 vintid); void vgic_v3_dispatch_sgi(struct kvm_vcpu *vcpu, u64 reg); diff --git a/virt/kvm/arm/arch_timer.c b/virt/kvm/arm/arch_timer.c index 70f4c30918eb..3945021510a9 100644 --- a/virt/kvm/arm/arch_timer.c +++ b/virt/kvm/arm/arch_timer.c @@ -581,6 +581,7 @@ void kvm_timer_sync_hwstate(struct kvm_vcpu *vcpu) int kvm_timer_vcpu_reset(struct kvm_vcpu *vcpu) { + struct arch_timer_cpu *timer = &vcpu->arch.timer_cpu; struct arch_timer_context *vtimer = vcpu_vtimer(vcpu); struct arch_timer_context *ptimer = vcpu_ptimer(vcpu); @@ -594,6 +595,9 @@ int kvm_timer_vcpu_reset(struct kvm_vcpu *vcpu) ptimer->cnt_ctl = 0; kvm_timer_update_state(vcpu); + if (timer->enabled && irqchip_in_kernel(vcpu->kvm)) + kvm_vgic_reset_mapped_irq(vcpu, vtimer->irq.irq); + return 0; } diff --git a/virt/kvm/arm/vgic/vgic.c b/virt/kvm/arm/vgic/vgic.c index c7c5ef190afa..0001858a2c23 100644 --- a/virt/kvm/arm/vgic/vgic.c +++ b/virt/kvm/arm/vgic/vgic.c @@ -495,6 +495,32 @@ int kvm_vgic_map_phys_irq(struct kvm_vcpu *vcpu, unsigned int host_irq, return ret; } +/** + * kvm_vgic_reset_mapped_irq - Reset a mapped IRQ + * @vcpu: The VCPU pointer + * @vintid: The INTID of the interrupt + * + * Reset the active and pending states of a mapped interrupt. Kernel + * subsystems injecting mapped interrupts should reset their interrupt lines + * when we are doing a reset of the VM. + */ +void kvm_vgic_reset_mapped_irq(struct kvm_vcpu *vcpu, u32 vintid) +{ + struct vgic_irq *irq = vgic_get_irq(vcpu->kvm, vcpu, vintid); + unsigned long flags; + + if (!irq->hw) + goto out; + + spin_lock_irqsave(&irq->irq_lock, flags); + irq->active = false; + irq->pending_latch = false; + irq->line_level = false; + spin_unlock_irqrestore(&irq->irq_lock, flags); +out: + vgic_put_irq(vcpu->kvm, irq); +} + int kvm_vgic_unmap_phys_irq(struct kvm_vcpu *vcpu, unsigned int vintid) { struct vgic_irq *irq;

7 years, 3 months

1
0
0 0

FAILED: patch "[PATCH] KVM: arm/arm64: Avoid vcpu_load for other vcpu ioctls than" failed to apply to 4.15-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.15-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From e21a4f3a930cda6e4902cb5b3213365e5ff3ce7c Mon Sep 17 00:00:00 2001 From: Christoffer Dall <christoffer.dall(a)linaro.org> Date: Tue, 27 Feb 2018 12:33:50 +0100 Subject: [PATCH] KVM: arm/arm64: Avoid vcpu_load for other vcpu ioctls than KVM_RUN Calling vcpu_load() registers preempt notifiers for this vcpu and calls kvm_arch_vcpu_load(). The latter will soon be doing a lot of heavy lifting on arm/arm64 and will try to do things such as enabling the virtual timer and setting us up to handle interrupts from the timer hardware. Loading state onto hardware registers and enabling hardware to signal interrupts can be problematic when we're not actually about to run the VCPU, because it makes it difficult to establish the right context when handling interrupts from the timer, and it makes the register access code difficult to reason about. Luckily, now when we call vcpu_load in each ioctl implementation, we can simply remove the call from the non-KVM_RUN vcpu ioctls, and our kvm_arch_vcpu_load() is only used for loading vcpu content to the physical CPU when we're actually going to run the vcpu. Cc: stable(a)vger.kernel.org Fixes: 9b062471e52a ("KVM: Move vcpu_load to arch-specific kvm_arch_vcpu_ioctl") Reviewed-by: Julien Grall <julien.grall(a)arm.com> Reviewed-by: Marc Zyngier <marc.zyngier(a)arm.com> Reviewed-by: Andrew Jones <drjones(a)redhat.com> Signed-off-by: Christoffer Dall <christoffer.dall(a)linaro.org> Signed-off-by: Marc Zyngier <marc.zyngier(a)arm.com> diff --git a/arch/arm64/kvm/guest.c b/arch/arm64/kvm/guest.c index d7e3299a7734..959e50d2588c 100644 --- a/arch/arm64/kvm/guest.c +++ b/arch/arm64/kvm/guest.c @@ -363,8 +363,6 @@ int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu *vcpu, { int ret = 0; - vcpu_load(vcpu); - trace_kvm_set_guest_debug(vcpu, dbg->control); if (dbg->control & ~KVM_GUESTDBG_VALID_MASK) { @@ -386,7 +384,6 @@ int kvm_arch_vcpu_ioctl_set_guest_debug(struct kvm_vcpu *vcpu, } out: - vcpu_put(vcpu); return ret; } diff --git a/virt/kvm/arm/arm.c b/virt/kvm/arm/arm.c index 86941f6181bb..53572304843b 100644 --- a/virt/kvm/arm/arm.c +++ b/virt/kvm/arm/arm.c @@ -384,14 +384,11 @@ static void vcpu_power_off(struct kvm_vcpu *vcpu) int kvm_arch_vcpu_ioctl_get_mpstate(struct kvm_vcpu *vcpu, struct kvm_mp_state *mp_state) { - vcpu_load(vcpu); - if (vcpu->arch.power_off) mp_state->mp_state = KVM_MP_STATE_STOPPED; else mp_state->mp_state = KVM_MP_STATE_RUNNABLE; - vcpu_put(vcpu); return 0; } @@ -400,8 +397,6 @@ int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu, { int ret = 0; - vcpu_load(vcpu); - switch (mp_state->mp_state) { case KVM_MP_STATE_RUNNABLE: vcpu->arch.power_off = false; @@ -413,7 +408,6 @@ int kvm_arch_vcpu_ioctl_set_mpstate(struct kvm_vcpu *vcpu, ret = -EINVAL; } - vcpu_put(vcpu); return ret; } @@ -1036,8 +1030,6 @@ long kvm_arch_vcpu_ioctl(struct file *filp, struct kvm_device_attr attr; long r; - vcpu_load(vcpu); - switch (ioctl) { case KVM_ARM_VCPU_INIT: { struct kvm_vcpu_init init; @@ -1114,7 +1106,6 @@ long kvm_arch_vcpu_ioctl(struct file *filp, r = -EINVAL; } - vcpu_put(vcpu); return r; }

7 years, 3 months

1
0
0 0

FAILED: patch "[PATCH] drm/i915/gvt: Correct the privilege shadow batch buffer" failed to apply to 4.15-stable tree

by gregkh＠linuxfoundation.org

The patch below does not apply to the 4.15-stable tree. If someone wants it applied there, or to any other stable or longterm tree, then please email the backport, including the original git commit id to <stable(a)vger.kernel.org>. thanks, greg k-h ------------------ original commit in Linus's tree ------------------ >From ef75c685869ea2059f85855a7dc00148a704c36c Mon Sep 17 00:00:00 2001 From: fred gao <fred.gao(a)intel.com> Date: Thu, 15 Mar 2018 13:21:10 +0800 Subject: [PATCH] drm/i915/gvt: Correct the privilege shadow batch buffer address Once the ring buffer is copied to ring_scan_buffer and scanned, the shadow batch buffer start address is only updated into ring_scan_buffer, not the real ring address allocated through intel_ring_begin in later copy_workload_to_ring_buffer. This patch is only to set the right shadow batch buffer address from Ring buffer, not include the shadow_wa_ctx. v2: - refine some comments. (Zhenyu) v3: - fix typo in title. (Zhenyu) v4: - remove the unnecessary comments. (Zhenyu) - add comments in bb_start_cmd_va update. (Zhenyu) Fixes: 0a53bc07f044 ("drm/i915/gvt: Separate cmd scan from request allocation") Cc: stable(a)vger.kernel.org # v4.15 Cc: Zhenyu Wang <zhenyuw(a)linux.intel.com> Cc: Yulei Zhang <yulei.zhang(a)intel.com> Signed-off-by: fred gao <fred.gao(a)intel.com> Signed-off-by: Zhenyu Wang <zhenyuw(a)linux.intel.com> diff --git a/drivers/gpu/drm/i915/gvt/cmd_parser.c b/drivers/gpu/drm/i915/gvt/cmd_parser.c index c8454ac43fae..db6b94dda5df 100644 --- a/drivers/gpu/drm/i915/gvt/cmd_parser.c +++ b/drivers/gpu/drm/i915/gvt/cmd_parser.c @@ -471,6 +471,7 @@ struct parser_exec_state { * used when ret from 2nd level batch buffer */ int saved_buf_addr_type; + bool is_ctx_wa; struct cmd_info *info; @@ -1715,6 +1716,11 @@ static int perform_bb_shadow(struct parser_exec_state *s) bb->accessing = true; bb->bb_start_cmd_va = s->ip_va; + if ((s->buf_type == BATCH_BUFFER_INSTRUCTION) && (!s->is_ctx_wa)) + bb->bb_offset = s->ip_va - s->rb_va; + else + bb->bb_offset = 0; + /* * ip_va saves the virtual address of the shadow batch buffer, while * ip_gma saves the graphics address of the original batch buffer. @@ -2571,6 +2577,7 @@ static int scan_workload(struct intel_vgpu_workload *workload) s.ring_tail = gma_tail; s.rb_va = workload->shadow_ring_buffer_va; s.workload = workload; + s.is_ctx_wa = false; if ((bypass_scan_mask & (1 << workload->ring_id)) || gma_head == gma_tail) @@ -2624,6 +2631,7 @@ static int scan_wa_ctx(struct intel_shadow_wa_ctx *wa_ctx) s.ring_tail = gma_tail; s.rb_va = wa_ctx->indirect_ctx.shadow_va; s.workload = workload; + s.is_ctx_wa = true; if (!intel_gvt_ggtt_validate_range(s.vgpu, s.ring_start, s.ring_size)) { ret = -EINVAL; diff --git a/drivers/gpu/drm/i915/gvt/scheduler.c b/drivers/gpu/drm/i915/gvt/scheduler.c index 8caf72c1e794..fdf1c0bf0d55 100644 --- a/drivers/gpu/drm/i915/gvt/scheduler.c +++ b/drivers/gpu/drm/i915/gvt/scheduler.c @@ -426,6 +426,17 @@ static int prepare_shadow_batch_buffer(struct intel_vgpu_workload *workload) goto err; } + /* For privilge batch buffer and not wa_ctx, the bb_start_cmd_va + * is only updated into ring_scan_buffer, not real ring address + * allocated in later copy_workload_to_ring_buffer. pls be noted + * shadow_ring_buffer_va is now pointed to real ring buffer va + * in copy_workload_to_ring_buffer. + */ + + if (bb->bb_offset) + bb->bb_start_cmd_va = workload->shadow_ring_buffer_va + + bb->bb_offset; + /* relocate shadow batch buffer */ bb->bb_start_cmd_va[1] = i915_ggtt_offset(bb->vma); if (gmadr_bytes == 8) diff --git a/drivers/gpu/drm/i915/gvt/scheduler.h b/drivers/gpu/drm/i915/gvt/scheduler.h index 2603336b7c6d..a79a4f60637e 100644 --- a/drivers/gpu/drm/i915/gvt/scheduler.h +++ b/drivers/gpu/drm/i915/gvt/scheduler.h @@ -124,6 +124,7 @@ struct intel_vgpu_shadow_bb { u32 *bb_start_cmd_va; unsigned int clflush; bool accessing; + unsigned long bb_offset; }; #define workload_q_head(vgpu, ring_id) \

7 years, 3 months

1
0
0 0

2025

2024

2023

2022

2021

2020

2019

2018

2017

Linux-stable-mirror